Version r.11.25 software procurve 2610 series (68 pages)
Summary of Contents for HP bh5700
Page 1
HP bh5700 ATCA 14-Slot Blade Server Ethernet Switch Blade First Edition Manufacturing Part Number: AD171-9603A June 2006 Downloaded from www.Manualslib.com manuals search engine...
Page 2
Ethernet Switch Blade User's Guide release 3.2.2j page ii Downloaded from www.Manualslib.com manuals search engine...
Legal Notices The information in this document is subject to change without notice. Hewlett-Packard makes no warranty of any kind with regard to this manual, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose.
Page 4
About the Ethernet Switch Blade Manual This manual includes everything you need to begin using the HP Ethernet Switch Blade with OpenArchitect software, Release 3.2.2j. Ethernet Switch Blade User's Guide release 3.2.2j page iv Downloaded from www.Manualslib.com manuals search engine...
Table of Contents Chapter 1 Overview of the Ethernet Switch Blade ............17 High Performance Embedded Switching..............17 Advanced TCA® Compliant..................17 OpenArchitect Switch Management................18 Extensible Customization of Routing Policies..............18 Powerful CarrierClass Features..................18 Ethernet Port Layout..................... 18 Ethernet Switch Blade Port Configuration..............19 Base switch Quick Reference................
Page 6
Rapid Spanning Tree....................50 To Enable Rapid Spanning Tree:.................51 Port Path Cost...................... 51 Layer 3 Switch Configuration................52 Using the S50layer3 Script.................. 52 Layer 3 Routing Protocols with GateD ................ 54 Using the S55gatedRip1 Script................54 To Modify the GateD Scripts: ................. 56 Class of Service (COS) ....................
Page 7
Chapter 5 Fabric Switch Administration................73 Setting the Root Password.................... 73 Adding Additional Users....................73 Setting up a Default Route.................... 74 Name Service Resolution....................74 DHCP Client Configuration..................74 DHCP Server Configuration..................74 Network Time Protocol (NTP) Client Configuration........... 75 Network File System (NFS) Client Configuration............75 NFS Server Configuration.....................76 Connecting to the Switch Using FTP................77 ftpd Server Configuration.....................
Page 8
Example Configuration Scripts................92 Overview of OpenArchitect VLAN Interfaces............93 Tagging and Untagging VLANs................94 Switch Port Interfaces..................94 Layer 2 Switch Configuration.................. 94 Using the S50layer2 Script.................. 96 Rapid Spanning Tree....................96 To Enable Rapid Spanning Tree:.................96 Port Path Cost...................... 97 Layer 3 Switch Configuration.................. 97 Using the S50layer3 Script..................
Page 9
Classical Targets....................111 ZNYX Targets....................112 ZACTION Examples..................112 Extensions to the default matches..............113 tc: Traffic Control....................113 Strict Priority Qdisc....................113 Weighted Round Robin Qdisc................114 FIFO Queues (pfifo and bfifo disciplines)............114 Fifo Qdiscs......................115 Using Filters to Direct Packets to a COS Queue............115 Protocol ip......................
Page 10
SNMP and OpenArchitect Interface Definitions..........134 ifStackTable Entries...................135 SNMP Configuration..................135 SNMP Applications................... 136 Port Mirroring......................136 Link and LED Control....................137 Link Event Monitoring................... 137 Chapter 9 Base Switch Maintenance................138 Overview of the OpenArchitect switch boot process..........138 Saving Changes...................... 140 Modifying Files and Updating the Switch..............140 Recovering from a System Failure.................
Page 11
Booting the Duplicate Flash Image ................159 Chapter 13 Network Configuration Problems ............... 160 Interface Overview......................160 Physical Interfaces....................160 Default Base Interface Configuration..............161 24 port, Layer 2 Switching, single VLAN............161 Default Fabric Interface Configuration..............163 Editing the S50layer2 script can change the Ethernet Switch Blade Fabric Interface default configuration.
Chapter 1 Overview of the Ethernet Switch Blade The Ethernet Switch Blade is a 72-port AdvancedTCA® Hub and providing Gigabit Ethernet. Up to 14 ATCA node boards may be addressed via the PICMG 3.0 Base Interface and via the ATCA PICMG 3.1 fabric . The Base and Fabric switching domains are kept totally separate, both on the physical layer and the software layer.
OpenArchitect Switch Management The OpenArchitect software component – open source Linux, IP protocol stack, control applications and the OA Engine – runs on two embedded PowerPC microprocessors. OpenArchitect provides extensive managed IP routing protocols and other open standards for switch management. Examples include network services; Virtual Redundant Router Protocol; Routing Information Protocol;...
You will find the Ethernet Switch Blade has a straightforward installation and configuration. UNIX or Linux system management skills and some understanding of network protocols will be required. Configure the Ethernet Switch Blades to your networking application before you begin using the OpenArchitect switch. OpenArchitect Switch Environment The key elements of the OpenArchitect environment include two embedded Linux operating systems, OpenArchitect-specific applications and libraries, plus, an innovative switch hardware...
Page 21
network-enabled Linux implementation. The purpose of the routing table is to tell the packet forwarding software where to forward the data packets. In Linux, the packet-forwarding algorithm is operated in software. Normally, the routing tables are maintained by operator configuration and the various routing protocols that run in the application environment of Linux.
The RS-232 configured RJ-45 connector console port on the front panel can be used to recover from a system failure. It is used for maintenance only, and is generally not connected. Use a HP console cable (P/N A6900-63006) provided with the HP bh5700 ATCA 14-Slot Blade Server, in combination with a Modem Eliminator cable, to access the switch software through the console port.
4. Reinsert the switch into the shelf chassis and power up. Use a terminal emulation program to access the switch console. Out of Band Ports (OOB Ports) Each switch, fabric and Base, in a Ethernet Switch Blade unit has out-of-band (OOB) Ethernet ports on the front panel.
Chapter 3 High Availability Networking High availability networking is achieved by eliminating any single point of failure through redundant connectivity: Redundant cables, switches and network interfaces for hardware, combined with HA software solutions on both the hosts and switches to control the HA hardware and maintain connectivity.
VRRP Since most end nodes use default router addresses, the change of the default router address during a switch failover would require the end nodes to reconfigure. Layer 3 switches that failover must maintain the default router address to maintain the end node's IP transparent failover. The Virtual Router Redundancy Protocol (VRRP, RFC 2338) running in the Surviving Partner switches provides transparent movement of the default router address.
Switch Replacement and Reconfiguration When a switch fails, it must be replaced. The replacement switch will likely require proper configuration. For transparent switch replacement, the newly replaced switch must learn its configuration from its Surviving Partner. In a simple failover scenario, Host A and Host B are configured with failover between two host ports, one port connected to Switch A and the other connected to Switch B.
The configuration and runtime scripts created are as follows: S70Surviving_partner Switch initialization script that is run at boot time. This • script will restart the switch with the original configuration given to zspconfig. Optionally, zspconfig will run this script from the initial invocation. zsp.conf.<n>...
When using a Linux Bonding driver on the node card, the bonding driver should be configured for Mode 1 (active/standby). See the Linux Bonding documentation at http://sourceforge.net/projects/bonding/ for complete information. The two Base switches will be configured as Surviving Partners, using VRRP to form a single virtual interface to the hosts, as will the two Fabric switches.
Page 32
sibling_addresses: zhp1 = 10.0.0.30, 10.0.0.31 netmask 255.0.0.0; Now configure the virtual address for each sibling group. We are going to create a virtual interface across one VLAN, but not for the interconnect. This provides a single point to connect/route to the VLANs. vrrp_virtual_address: zhp1 = 10.0.0.42 netmask 255.0.0.0;...
Page 33
#vrrp_mode: block_crossconnect; The next sections determines the failover mode between the Surviving Partner switches. There are three modes: switch - Failover by switch. Failover from Master switch to Backup on any port • failure. The switch with the most links becomes the new Master. One port failure will cause the switch to failover.
Page 34
#start_script:/etc/rcZ.d/SxxScript; #start_script:/etc/rcZ.d/SyyScript; # vrrpd_script: Allows the user to add scripts to be executed during # vrrpd state transitions. These scripts are run from the end of the # /etc/rcZ.d/surviving_partner/vrrpd.script file. The user provided # script must be well behaved. If it crashes, or hangs or delays it will # effect the SurvivingPartner performance.
Once the configuration files are complete, run the zspconfig utility on the Master to configure all the scripts: NOTE: This command can take 60 seconds or more with no screen output. zspconfig –f zsp.conf You will see output similar to this: zspconfig -f zsp.conf ….
Page 36
# This script will likely need modification for your particular # network setup. # In this example the Egress ports, zre20..23 and zre48..50 are # not managed by HA since how, or if, these ports are managed by HA is # dependent on the external devices they are connected to.
Page 38
vrrp_virtual_address: zhp1 = 10.0.0.42 netmask 255.0.0.0; vrrp_virtual_address: zhp2 = 11.0.0.42 netmask 255.0.0.0; vrrp_virtual_address: zhp3 = 12.0.0.42 netmask 255.0.0.0; vrrp_virtual_address: zhp4 = 13.0.0.42 netmask 255.0.0.0; # Port definitions # Define to what the ports are connected. Specifications can # by zhp or zre name. The zhp name is a shortcut to specify # entire port group associated with that interface.
Page 39
# crossconnect ports of the VRRP Backup. block_crossconnect mode is # meant as a replacement for STP, however, the switches connected to the # crossconnect ports must be Ethernet Switch switches running Surviving Partner. # The RAINlink_xmit_on_failover mode requires that the OpenNode blades # connected to RAINlink ports transmit a packet when failing over, so that...
Page 40
failover_mode: port; # VRRP_msg_rate is the time in milliseconds between transmissions # VRRP messages on the interconnect. The VRRP protocol requires the # absence of 3 VRRP messages before concluding that the remote switch # has failed. The msg_rate must match the msg_rate of all siblings.
Page 41
# Fabric portions of the 7100 switch. The actual coordination is dependent on the # setting of the board_synchronization_mode and the failover_mode. # switch failover_mode the number of up links in both switch planes is # considered. In vlan and port failover mode they are not. # failover_modes, if the data plane or fabric plane switch reboots or # power cycles, the HA partner will take mastership for all...
# gated_template: Allows the user to provide a template for # gated.conf file to be used by the sibling group. #gated_template: /etc/rcZ.d/surviving_partner/gated.template Once the configuration files are complete, run the zspconfig utility on the Master to configure all the scripts: NOTE: This command can take 60 seconds or more with no screen output.
Finally, it lets the currently saved S70Surviving_Partner script execute. This case would be the case of a power up of an already configured backup switch when the other HA switch is unavailable. This case could occur after losing power to the entire chassis. Central Authority Modifications can be made to the S60SP_startup script to use a third machine running DHCP that is not part of the Surviving Partner pair.
Page 44
"zsp.primary.conf"; host SECONDARY { fixed-address 100.0.0.31; option dhcp-client-identifier "SECONDARY"; option vendor-encapsulated-options "zsp.secondary.conf"; The zsp.primary.conf and zsp.secondary.conf files must be placed in the tftp location on the machine, often /tftpboot. The zsp.primary.conf and zsp.secondary.conf files can be retrieved from the Surviving Partner switches. This is the configuration that will be given to the switches.
Page 45
request vendor-encapsulated-options; require vendor-encapsulated-options; The last step is to modify the startup scripts that run zspconfig to use the -c option. The -c option allows you to provide a dhclient.conf script rather then having zspconfig create a default. For example, the S60SP_startup script line that reads: echo y n | zspconfig -t 10 -su zhp0 >...
Chapter 4 Fabric Switch Configuration Two switches, two consoles There are two separate switch portions in the Ethernet Switch Blade units, the base switch and the fabric switch. The fabric switch handles the data traffic for the ATCA rack over ports 0-47. It runs the Ethernet Switch Blade software.
Changing the Shell Prompt You may use standard bash shell procedures to change the prompts on your base switches. Many sites choose a system that distinguishes among the individual switches at their location. The same rules apply for saving your choice (zsync) as for all other configuration changes. Default Configuration Scripts As shipped the following scripts are run from as the switch boots up:...
Overview of OpenArchitect VLAN Interfaces A zhp device is associated with one VLAN. zhp may have one or more physical ports and their associated zre devices. A VLAN from the viewpoint of the switch is a logical mapping of ports based on intended use.
Switch Port Interfaces For each switch port, OpenArchitect creates a separate interface with its own MAC address called a ZNYX raw Ethernet (zre). After the initial power up, 48 zre interfaces are created, one for each in band port. You cannot directly access or modify the zre interfaces. During the initial power up of the switch, the default configuration creates a Layer 2 switch.
ifconfig zhp1 0.0.0.0 # At this point the system will act as a Layer 2 switch # across all ports. Also, the system will accept telnet() # connections on 10.0.0.43 on any port. Script(s) may then # be run to reinitialize the system and modify its # configuration.
To Enable Rapid Spanning Tree: Create a VLAN containing the ports that will be a part of the Linux bridge running Rapid Spanning Tree. This example will use ports 0-3 (untagged): zconfig zhp0: vlan1=zre0..3 zconfig zre0..3=untag1 Create a bridge device from the zhp device, zl2d start zhp0 A Bridge device named bzhp0 should now exist consisting of ports zre0 through zre3 with Spanning Tree enabled.
Layer 3 Switch Configuration The previous section outlines the Layer 2 switch configuration that is automatically configured when you initially bring up the OpenArchitect switch. In order to communicate between Layer2 interfaces, you must properly setup routing. The steps to build a Layer 2 switch involve creating a group of switch ports in a VLAN (or Layer 2 switching domain) and bringing that interface up.
Page 53
In the S50layer3 script separate VLANs are set up for each port. The VLANs, are labeled as zhp0..zhpn. Each VLAN is associated with an individual zre interface. There is always a one to one connection between VLANs and zhp interfaces. Remember, zre and zhp interfaces can begin with a zero value but a VLAN cannot (that is, zhp0 has zre0 on vlan1, zhp1 has zre1 on vlan2).
the number of IP addresses as applicable. In the example below, the IP address is changed for the interface in the ifconfig command line of the script. From: ifconfig zhp0 10.0.0.43 netmask 255.255.255.0 broadcast 10.0.0.255 up ifconfig zhp0 193.08.1.1 netmask 255.255.255.0 broadcast 193.08.1.255 up Adjust the number of zhp interfaces, that are added to the routing tables, depending on the •...
interface 10.0.13.43 ripin ripout version 1; interface 10.0.14.43 ripin ripout version 1; interface 10.0.15.43 ripin ripout version 1; Imports routes learned through the RIP protocol. • import proto rip { all; Exports all directly connected routes and routes learned from the RIP protocol. •...
Or for OSPF: cp /etc/rcZ.d/examples/S55gatedOspf /etc/rcZ.d cp /etc/rcZ.d/examples/gated.conf.ospf /etc/rcZ.d Open and make configuration changes to the listed conf file to coincide with the current • Layer 3 configuration (that is, adjust IP addresses and number of interfaces available). See GateD documentation if you have questions regarding the conf file. Run the OpenArchitect zsync command to save your changes.
Marking and Re-marking The OpenArchitect switch can mark or remark packets using the TOS field or 802.1p tag. This is also controlled through the Linux iptables utility. Scheduling The servicing of configured queues by the switching fabric is referred to as scheduling. The OpenArchitect switch has three built-in scheduling algorithms.
you may want to move your set of iptables commands to a start up script to run upon initialization. This could be accomplished by creating a standalone "S" script and placing that script into / etc/rcZ.d. Restrictions on Implementation Several restrictions exist on the rules that can be implemented on the FFP hardware. These include: Actions DROP the packet.
On the other hand, in the following sequence of rules, the position of the rule that drops SYN packets is important. Since the set of fields it examines is not a subset of the fields examined by the ACCEPT rules, and visa versa, the ordering rule given above does not apply. In this case, the order it is applied will be the same as its position in the FORWARD chain, and all packets which are TCP SYN packets from zhp5 for zhp3 will be DROPPED, even if they also match one of the ACCEPT rules.
By default, INPUT, FORWARD and OUTPUT chains are installed on boot up. Additional rules can be installed for the other chains. Additionally, one can write software extensions to add more chains. Figure 4.2 provides an illustration of the Firewall Flow. R o u t i n g P o s t F o r w a r d...
send to CPU action is specified, it is sent to the INPUT chain for further processing. If there is no valid way to forward the packet, it is dropped. If the switch is configured to forward the packet, it is sent to the FORWARD chain. Next the hardware FORWARD chain is walked.
The type can be preceded by ! to match any message except the type listed, for example, -- icmp-type ! 1 Specifying TCP or UDP ports If the protocol is TCP or UDP, the -s ( or --sport) and -d (or --dport) options specify the TCP or UDP ports to match.
--drop Drops the packet --accept Accepts the packet --set-prio <val>Set the 802.1p priority to <val> --use-prio <val>Use queue priority <val> --copy-cpu Send the packet to the CPU. This will force the full installed chains traversal in software --set-eport <val> Redirect the packet to port <val> --set-mport <val>...
FORWARDING Chain supports all of them. tc and zqosd tc, which stands for Traffic Control, is a mechanism for enabling Quality of Service on Linux. tc uses three functional objects: queuing disciplines, which comprise queuing and scheduling algorithms such as FIFO queues, priority queues, RED queues, and token buckets; classes, which are leafs in queuing discipline hierarchies;...
Page 66
qdisc pfifo 100: dev zhp0 limit 32p The tc command is applied to a device, so dev zhp0 must be specified. Note that a VLAN, such as zhp0, and a port, such as zre0, are each treated as devices. Breakdown of the options: handle 100:0 Defines the handle for the queuing discipline.
The byte-limited FIFO queue case differs only slightly from the packet-limited FIFO case. The syntax is almost identical. In hardware the limit is based on 128-byte cells. The specified byte limit is divided by 128 to determine the cell limit. Always specify a byte limit of at least 128 bytes to avoid setting the queue length to zero.
Page 68
index of the list element (numbering from 0) and q is the value specified by that element. So, this example would read: Priority 0 maps to Queue 1 Priority 1 maps to Queue 2 Priority 2 maps to Queue 2 Priority 3 maps to Queue 2 Priority 4 maps to Queue 3 Note that the tc priority map applies to a 4-bit field.
The U32 Filter The U32 filter provides the capability to match on fields in the L2, L3 or L4 header of a packet. Each match rule gives the location of the field to be tested, which is always a 32 bit word, a mask selecting the bits to be tested, and a value which is to be matched by the packet field.
Although the translation rules handle some inconsistency between software and hardware, a user must define a combination of rules that is reasonable in hardware, to ensure predictable results. Handle Semantics All examples have illustrated zqosd copying tc rules into hardware. In fact, the zqosd utility also enables the user to add tc rules that remain only in software.
The PDP sends that policy to the PEP. • The PEP installs the policy and applies it to future traffic. • As long as COPS is running, a connection between the PEP and PDP should stay open. A PEP could query a PDP at any time asking for a policy decision. Alternatively, an administrator could modify the policy on a PDP, which would then push any policy changes to its PEPs.
The pepd utility requires a PDP that has implemented the above RFCs and drafts. Until all draft standards are approved, the certain COPS-PR data types will not be assigned OIDs. pepd uses non-standard OIDs for the unassigned values. Using pepd The pepd utility works by connection to a PDP, informing the PDP of its roles, and installing any rules that the PDP has for those roles.
Chapter 5 Fabric Switch Administration One of the main benefits of the OpenArchitect switch is that it runs Linux, so much of the switch administration is already familiar to most network or system administrators. It is a good idea to complement these instructions with a standard Linux reference guide, such as Linux Network Administrator’s Guide available from O’Reilly.
Enter new password: Re-enter new password: Password changed. ZX7100-OA<release no.># zsync ZX7100-OA<release no.># Setting up a Default Route If you wish to access the switch from some place other than a directly attached network, you may want to setup a default route. Use the route command to set a default gateway. route add default gw 10.0.0.254 Put the entry into the /etc/init.d/rcS startup script to automatically set a default route upon reboot.
dhcpd Consult Linux Network administration manuals for more information on DHCP and configuration options. To use DHCP to set your IP addresses automatically on boot up, uncomment the the following line in /etc/init.d/rcS by removing the # sign dhcpd Network Time Protocol (NTP) Client Configuration NTP is a protocol for setting the real time clock on a system.
/sbin/rpc.statd /usr/sbin/rpc.mountd -r Once the above servers are started, you can mount a remote NFS file system. mount rhost:nfs_file_system local_mount_point If the remote NFS file system you’re mounting is on an OA switch, you should mount with caching disabled. mount rhost:nfs_file_system –o noac local_mount_point All the necessary servers are included in /etc/init.d/rcS but are commented out by default.
Now start nfsd to export the mount points and begin answering requests from remote clients. /sbin/rpc.nfsd –r To export file systems automatically on boot, edit /etc/init.d/rcS, uncomment the /sbin/rpc.nfsd command line by removing the #. /sbin/rpc.nfsd -r Connecting to the Switch Using FTP Use ftp to transfer files to or from the switch.
SNMP Agent Simple Network Management Protocol (SNMP) is the defacto standard for network management. An SNMP agent maintains a structure of data for a network device in a virtual information database, called a Management Information Base (MIB). A network management station is capable of accessing the MIB of the network device to monitor and configure the network device.
Supported MIBs RFC 2573: SNMP Applications RFC 2574: User-based Security Model (USM) for version 3 of the Simple Network Management Protocol (SNMPv3) RFC 2575: View-based Security Model (VACM) for version 3 of the Simple Network Management Protocol (SNMP) RFC 2576: Coexistence between Version 1, Version 2 and Version 3 of the Internet- standard Network Management Framework RFC 2665:...
Link and SNMP Status Physical Link Status SNMP Operational Status zhp0 zre1 zre2 zre1 zre2 down down down down down down down down down Table 5.3: Link and SNMP Status The administrative status is directly controlled by ifconfig up/down. The administrative status of the zhps and zres do not affect each other.
response. The processing for link up and link down traps is now user configurable. As the default, traps conform to RFC2863, meaning the trap contents will include: ifIndex, ifAdminStatus and ifOperstatus You can alter this behavior by specifying: cisco_link_traps on If cisco_link_traps are turned on as described then link up and link down traps will have a cisco-like format and the trap contents will include: ifDescr and ifType...
mirrored (copied and transmitted) to port 12. This mirroring would be in addition to any Layer 3 or Layer 2 switching. zmirror zre0 zre12 zmirror zre1 zre12 zmirror zre2 zre12 To clear the current mirroring use the -t option. The -e option can be used to indicate that packets being sent on a given port should be copied to the mirror_to port.
Chapter 6 Fabric Switch Maintenance This chapter includes basic information about the OpenArchitect switch environment including an overview of the file system structure, modifying and updating switch files, upgrading the switch driver and kernel, and implementing a system recovery. Overview of the OpenArchitect switch boot process The OpenArchitect switch is equipped with a Random Access Memory (RAM) disk and three Read-Only Memory (ROM) devices, including, a boot ROM and two application flash devices.
Bootloader examines the bootstring in the boot Determines Loads image if the boot string from Flash 1 is dev1 to RAM Determines Loads image if the boot String from Flash 2 is dev2 to RAM Begins Boot into execution of zmon bootloader RAM image Figure 6.2: Boot Flow Chart...
/etc/init.d/rcS /etc/rcZ.d/rc Figure 6.3: Init Script Flow Saving Changes Any modifications made to the scripts for your particular configuration must be properly saved or your changes are lost when you reboot. The file system for the switch only exists in memory. A rewritable overlay is contained within the upper four megabytes of the first application flash.
configuration files contained in /e t c / r c Z . d In order to telnet into the box, there must be a configured interface with a proper IP address. For example, zhp0 is configured with the IP address 10.0.0.43 in the factory default configuration. Booting with the –i option If you cannot telnet into the switch and Linux fails to boot, it is likely that a change saved by zsync has left the switch in an inaccessible state.
zsync /etc/hosts Reboot the system. • System Hangs During Boot After attaching the system console cable, if the system hangs during boot, try booting with the –i option as described in the previous section. It is possible that important Linux system files became corrupted and incorrectly saved in the flash overlay.
Download the OpenArchitect image to a local system. The OpenArchitect image is very close to the limit of free space available on a default system so you may need to clear some space prior to downloading the OpenArchitect image to the switch. Check for free space with the df command.
Using apt-get apt-get is a utility created by the Debian Linux community to allow remote fetching and installation of software stored in a repository in Debian package format. It allows users to keep their software up-to-date with the latest binaries, and install new software without the need to recompile.
Chapter 7 Base Switch Configuration At this point, the OpenArchitect Ethernet Switch Blade should be installed and powered up for the first time. This chapter helps you connect and configure the base switch by presenting command line examples as well as a discussion of the example configuration scripts. You may configure the fabric switch independently from the base switch.
files into flash for reloading. Changing the Shell Prompt You may use standard bash shell procedures to change the prompts on your base switches. Many sites choose a system that distinguishes among the individual switches at their location. The same rules apply for saving your choice (zsync) as for all other configuration changes. Default Configuration Scripts As shipped the following scripts are run from as the switch boots up:...
S50multivlan - Script which sets up multiple untagged VLANs. The first VLAN • includes the first ten 10/100/1000 ports, the next contains the last ten 10/100/1000 ports, the third VLAN contains two 10/100/1000 ports, the last VLAN contains the last two 10/100/1000 ports. Layer 3 switching is enabled. S55gatedRip1 - Script which is used with a Layer 3 switch and calls the GateD •...
Figure 7.1: Multiple VLANs Tagging and Untagging VLANs The OpenArchitect switch is capable of switching VLAN tagged and untagged data packets. VLAN tagged packets conform to the 802.1q specification and the packet header contains an additional four bytes of VLAN tag information. A given port can be specified to accept VLAN tagged or untagged traffic.
Linux IP zhp0 10.0.0.42 VLAN 1 ..zre0 zre1 zre2 zre20 zre22 zre23 24 10/100/1000 Ports Figure 7.2: Layer 2 Switch During the initial power up, a startup script called /etc/rcZ.d/S50layer2 is executed at boot time creating a single untagged VLAN (IP interface labeled as zhp0) which includes all Ethernet and gigabit ports as one Layer2 switch.
Using the S50layer2 Script The S50layer2 script can be used and example, or edited to customize your Layer2 setup. For example, to reconfigure the IP address on your Layer 2 switch, Open the S50Layer2 file in the Linux vi editor. •...
brctl show brctl showbr bzhp0 Port Path Cost Each port has an associated cost that contributes to the total cost of the path to the Root Bridge when the port is the root port. The smaller the cost, the better the path. The Ethernet Switch Blade uses the following IEEE 802.1D recommendations based on the connection speed of your port: Port Path Cost Link...
zconfig zhp1: vlan2=zre5..8 zconfig zre5..8=untag2 Now, use ifconfig to assign each zhp interface an IP address, ifconfig zhp0 10.0.0.1 ifconfig zhp1 11.0.0.1 At this point, the Linux host has enough information to route between the networks of the directly attached interfaces, 10.0.0.0 via zhp0, and 11.0.0.0 via zhp1. The next step is to enable the ZNYX zl3d daemon to move that routing information from the host to the base switch switching tables in silicon.
Runs the OpenArchitect zl3d. The zl3d application monitors the Linux routing tables • and updates the switch routing tables for each interface configured above. /usr/sbin/zl3d zhp0..23 zl3d initially creates and adds each zhp interface (VLAN) to the switch routing tables. The zhp0..zhp23 is shorthand for the list of interfaces (zhp0, zhp1, …, zhp23) to monitor with zl3d.
VLAN 4, zhp3: for last set of six ports, zre18-zre23 • Each VLAN interface is labeled zh p N in the file, where N is a value from 0-3. Each interface is untagged and assigned its own IP address (see Figure 7.4). Linux IP zhp0 zhp1...
(10.0.0.42-10.0.3.42), assigns the netmask and brings them up. ifconfig zhp0 10.0.0.42 netmask 255.255.255.0 broadcast 10.0.0.255 up ifconfig zhp1 10.0.1.42 netmask 255.255.255.0 broadcast 10.0.1.255 up ifconfig zhp2 10.0.2.42 netmask 255.255.255.0 broadcast 10.0.2.255 up ifconfig zhp3 10.0.3.42 netmask 255.255.255.0 broadcast 10.0.3.255 up Runs the OpenArchitect zl3d command.
Page 103
example): Starts GateD with Rip1 using gated.conf.rip1 as the configuration file: • /usr/sbin/gated –f /etc/rcZ.d/gated.conf.rip1 The GateD conf file specifies the following configuration commands: Implements the passive function so GateD is prevented from rerouting information to a • different interface if insufficient information is received. interface 10.0.0.42 passive interface 10.0.1.42 passive interface 10.0.2.42 passive...
cp /etc/rcZ.d/examples/gated.conf.rip1 /etc/rcZ.d Or for RIP2: cp /etc/rcZ.d/examples/S55gatedRip2 /etc/rcZ.d cp /etc/rcZ.d/examples/gated.conf.rip2 /etc/rcZ.d Or for OSPF: cp /etc/rcZ.d/examples/S55gatedOspf /etc/rcZ.d cp /etc/rcZ.d/examples/gated.conf.ospf /etc/rcZ.d Open and make configuration changes to the listed co n f file to coincide with the current • Layer 3 configuration (that is, adjust IP addresses and number of interfaces available). See GateD documentation if you have questions regarding the co n f file.
Marking and Re-marking The OpenArchitect switch can mark or remark packets using the TOS field or 802.1p tag. This is also controlled through the Linux iptables utility. Scheduling The servicing of configured queues by the switching fabric is referred to as scheduling. The OpenArchitect switch has three built-in scheduling algorithms.
Running zfilterd Before starting zfilterd, ztmd must be running. Your can start both from within a script, or directly from the command line. For example, ztmd zfilterd iptables rules can be entered at any time. If your iptables filtering rules set is extensive, you may want to move your set of iptables commands to a start up script to run upon initialization.
action that will take place. For example, the rules: iptables -a FORWARD -i zhp3 -j DROP iptables -a FORWARD -i zhp3 -o zhp1 -p tcp --dport smtp -j ACCEPT result in SMTP packets received on any port in zhp3 to be sent for any port in zhp1; all other packets from zhp3 would be dropped.
Introduction Firewall rules are stored in tables. These tables are sometimes also known as firewall chains or just chains. Tables normally store rules for what are known as hooks, which can be looked as packet-path junctions. There are five defined hooks: PRE-ROUTE, POST-ROUTE, INPUT, OUTPUT and FORWARDING.
Packet Walk When a packet comes in via one of the interface ports, the base switch makes a routing decision. If the packet was destined for the base switch itself or if the send to CPU action is specified, it is sent to the INPUT chain for further processing.
Supported Targets The following are the supported targets. The switch has many additional targets that are software based (example Network Address Translation or generic connection tracking). Please contact HP Technical support if you have additional questions on additional features. Classical Targets DROP This drops the packet.
ZNYX Targets ZACTION This is the ZNYX Action target. Parameters for ZACTION: --drop Drops the packet --accept Accepts the packet --set-prio <val> Set the 802.1p priority to <val> --use-prio <val> Use queue priority <val> --copy-cpu Send the packet to the CPU. This will force the full installed chains traversal in software --set-eport <val>...
Extensions to the default matches These are described in the Linux packet filtering HOWTO at: http://netfilter.org/documentation/index.html#documentation-howto ZNYX FORWARDING Chain supports all of them. tc: Traffic Control The switch supports up to eight queues for each port, including the cpu port. These queues hold packets waiting to be transmitted for a given port.
queue number + 1 after the qdisc handle. So the highest priority queue in this example is 105:8. NOTE: 16 values must be provided for the priomap list. This is a feature of the Linux priority system, which uses 16 priority levels. The last eight values given will be ignored. Weighted Round Robin Qdisc A weighted round robin qdisc builds on the above definition by adding the list of weights which determine the order of scheduling from the queues:...
handle 100:0 Defines the handle for the queuing discipline. This handle may be used to reference the pfifo queue. Note that the handle is included with the output of the qdisc ls command. (100:0 and 100: are equivalent in tc.) The choice of handle is significant for zqosd.
match ip tos 0xa0 0xe0 would match an IP precedence of 5. Specific fields can also be specified by giving their offset from the beginning of the IP header and a field name of u8, u16, or u32, depending on the width of the field. For example, to match the SYN bit in the TCP flags, the specification is: match u8 2 0x02 at 33 Several IP fields can be matched in the same filter by specifying multiple match operations.
tc qdisc add dev zre1 ingress //ingress qdisc for zre1 tc qdisc add dev zhp2 ingress //ingress qdisc for vlan The filter add command changes slightly, the parent is now a special handle ffff:fff1, so using the same filter as the first example: tc filter add dev zre1 parent ffff:fff1 protocol ip u32 match ip dst 10.91.100.5/32 classid 105:2 This filter will match packets arriving on port zre1, destined for port zre5, with destination IP...
omitted, and the packet is not dropped, the egress queue will be determined by the priority of the packet, either from the 802.1p priority for tagged packets or the default priority for untagged packets for the ingress port. Examples The following commands set up priority queues for packets sent to the CPU and then use filters with policing to direct packets into these queues and limit their bandwidth.
specified numerically for either out-of-profile or in-profile actions. The numeric value is a decimal integer action code shown in the table below. If the action requires a parameter, the parameter value is multiplied by 256 and added to the action code. Only a few of the actions are possible for out-of-profile.
for a u16 match. In many cases, there is a field name that can be used for the match, eliminating the need to specify the offset. U match selectors Field Match Equivalent ip src a.b.c.d/n u32 <value> <mask> at 12 ip dst a.b.c.d/n u32 <value>...
OpenArchitect switch though, because the normal case is for packets to be switched in hardware. For that reason, zqosd must be used to shadow tc configuration into hardware. Like zfilterd, zqosd works with ztmd, which provides the actual hardware interaction. If ztmd is not already running, start it:, then initiate the zqosd daemon with no parameters: ztmd zqosd...
Page 122
In tc, the prio queuing discipline establishes multiple queues and specifies their associated priority map. Although WRR support is not part of the standard tc distribution, it has been added to the prio disciplinE. The following example illustrates WRR. A strict priority scheduler is a simpler case that can be constructed easily from this example.
many packets sent as queue 0, queue 2 will have four times as many, and queue 3 will have six times as many. wrr parameters are scaled such that the maximum value is no more than 15. values which would be 0 are set to 1: Queue 0 has a weight of 1000 bytes •...
tc filter add dev zhp0 protocol arp parent 100:0 u32 match u32 2 0xffff at +4 flowid 100:30 Combining Queuing Disciplines Any of the queue length limiting disciplines can be used with the bandwidth management queue disciplines, by defining them with the handle of one of the classes as their parent. For the htb queueing discipline, each class has an explicit handle specified when it is defined.
Figure 7.6: COPS Network Architecture A PDP contains all of the policy rulers for its associated PEPs. A PDP typically stores rules in a data and is a dedicated server, not a forwarding device. A PEP is any network device that has to enforce policy decisions. For example, a switch that restricts network access or prioritizes traffic fits the definition of a Policy Enforcement Point.
and relaying those requests to its PDP. By contrast, the provisioning model is based on longer lasting policy. The expectation is that policy should be administratively defined at the PDP and pushed to the PEPs as needed. OpenArchitect is a COPS-PR client. The most common use of COPS-PR is for distributing Differentiated Services (Diffserv) policy.
Page 127
where, PDP address: The IP address of the PDP. Default is loopback (127.0.0.1) PDP port: The destination port on which to open a COPS connection. Default is 3288. PEPID: The PEP Identifier Role-If: A mapping of roles to interfaces. The name of the role is followed by a comma-delineated list of interfaces.
Chapter 8 Base Switch Administration One of the main benefits of the OpenArchitect switch is that it runs Linux, so much of the switch administration is already familiar to most network or system administrators. It is a good idea to complement these instructions with a standard Linux reference guide, such as Linux Network Administrator’s Guide available from O’Reilly.
ZX6000-OA<release no.># zsync ZX6000-OA<release no.># Setting up a Default Route If you wish to access the switch from some place other than a directly attached network, you may want to setup a default route. Use the route command to set a default gateway. route add default gw 10.0.0.254 Put the entry into the /etc/init.d/rcS startup script to automatically set a default route upon reboot.
Network Time Protocol (NTP) Client Configuration NTP is a protocol for setting the real time clock on a system. There are numerous primary and secondary servers available on the network. For more NTP information, and a list of available NTP servers, see the following URL: http://www.ntp.org/ You will need to have your network settings properly configured to reach an available NTP server on your local network or the Internet.
All the necessary servers are included in /etc/init.d/rcS but are commented out by default. To automatically start all NFS client services each time you boot, uncomment the NFS Client servers. Go to the /etc/init.d/rcS file. Uncomment the following command lines by removing the # sign.
ftpd Server Configuration The switch itself can also be configured to run an FTP server (ftpd). See the Linux Reference Guide for details of the ftpd command. You will need to add a user to the switch in order to connect via ftp from a remote host, since root is not allowed ftp access.
Page 133
Supported MIBS RFC 1155: Structure and Identification of Management Information for TCP/IP-based internets RFC 1227: SNMP MUX Protocol and MIB RFC 1493: Definitions of Managed Objects for Bridges (obsoletes RFC 1286) RFC 1657: Definitions of Managed Objects for the Fourth Version of the Border Gateway Protocol (BGP-4) using SMI-V2 RFC 1724: RIP Version 2 MIB Extension (obsoletes RFC 1389)
Supported MIBS ZNYX Networks Private Custom ZNYX MIB to support software and hardware features not covered by standard MIBs. The Private MIBs are ZX7100BASE.MIB AND ZX7100FABRIC.MIB, pointed to by ZNYX-H.MIB. UCD-SNMP Enterprise UCD-SNMP MIB related to management and monitoring of the LINUX host Table 8.1: Supported MIBs Supported Traps Upon certain events, the OpenArchitect switch can be configured to send notification of the...
status is down, then the operational status will be down independent of the underlying link state. You must ifconfig up the zres to see the operational link status for a zre. When the administrative status is up, the operational status is dependent on the underlying physical state. For example, Table 8.3 shows that if zhp0 contains zre1 and zre2, the it would also be true for the operational status (given the administrative status is up on zre1, zre2, and zhp0).
IMPORTANT: For NET-SNMP agents, these objects (sysLocation.0, sysContact.0 and sysName.0) ordinarily are read-write. However, specifying the value for one of these objects by giving the appropriate token in snmpd.conf makes the corresponding object read-only, and attempts to set the value of the object will result in a notWritable error response.
zmirror mirror_from mirror_to After executing the following three commands, packets received on ports 0, 1 and 2 would be mirrored (copied and transmitted) to port 12. This mirroring would be in addition to any Layer 3 or Layer 2 switching. zmirror zre0 zre12 zmirror zre1 zre12 zmirror zre2 zre12...
Chapter 9 Base Switch Maintenance This chapter includes basic information about the OpenArchitect switch environment including an overview of the file system structure, modifying and updating switch files, upgrading the switch driver and kernel, and implementing a system recovery. Overview of the OpenArchitect switch boot process The OpenArchitect switch is equipped with a Random Access Memory (RAM) disk and three Read Only Memory (ROM) devices, including, a boot ROM and two application flash Application...
Bootloader examines the bootstring in the boot Determines Loads image if the boot string from Flash 1 is dev1 to RAM Determines Loads image if the boot String from Flash 2 is dev2 to RAM Begins Boot into execution of zmon bootloader RAM image Figure 9.2: Booting up Process Flow...
/etc/init.d/rcS /etc/rcZ.d/rc Figure 9.3: Init Script Flow Saving Changes Any modifications made to the scripts for your particular configuration must be properly saved or your changes are lost when you reboot. The file system for the switch only exists in memory. A rewritable overlay is contained within the upper four megabytes of the first application flash.
Booting with the –i option If you cannot telnet into the switch and Linux fails to boot, it is likely that a change saved by zsync has left the switch in an inaccessible state. To allow users to recover from mistakes saved in the overlay file system, a boot argument of –i passed to the init process will stop the untarring of the saved overlay files.
–d 1 rdr6000.zImage.initrd Upgrading the OpenArchitect Image 1. Refer to the HP bh5700 14-Slot Blade Server Installation Guide, Chapter 6, 14-Slot Shelf Startup, Validating and Updating Your Firmware for instructions on how to gain access to firmware updates for the HP bh5700 14-Slot Blade Server.
the limit of free space available on a default system, so you may need to clear some space prior to downloading the new OpenArchitect image to the switch. CAUTION: Do not remove the existing copy of /usr/sbin/gated (as suggested in Step 5, below) until you have, in fact, determined that an OpenArchitect upgrade version is available for downloading.
Using apt-get apt-get is a utility created by the Debian Linux community to allow remote fetching and installation of software stored in a repository in Debian package format. It allows users to keep their software up-to-date with the latest binaries, and install new software without the need to recompile.
Chapter 10 Connecting to the Ethernet Switch Blade The Ethernet Switch Blade has two completely separate switching subsystems within one ATCA blade supporting both Base Interface and Fabric Interfaces Figure 10.1: Fabric and Base The Ethernet Switch Blade implements an independent control processor and software environment for both Base and Fabric Interface switching subsystems.
console port. An RS-232 to RJ-45 adapter is required. Fabric Interface Hub System A 48-port Gigabit Ethernet Switch that provides PICMG 3.1 Option 2 (2.0 Gb/s) Ethernet service for a full 14-slot ATCA chassis. All connectors for the fabric interface hub and it’s processor are labeled “fabric”.
To attach the console cable to the Ethernet Switch Blade switch: 1. Plug the RJ-45 end of the console cable (P/N 6900-63006, shipped with the HP bh5700 ATCA 14-Slot Blade Server) into the RJ-45 Console Port (1) on the front panel.
To attach the console cable to the Ethernet Switch Blade switch: 6. Plug the RJ-45 end of the console cable (P/N 6900-63006, shipped with the HP by5700 ATCA 14-Slot Blade Server) into the RJ-45 Console Port (1) on the front panel.
9. Reinsert the switch into the system and power up. 10. Use a terminal emulation program to access the switch console. Fabric Interface Out of Band Ethernet Connection Connect an Ethernet cable from the Ethernet Switch Blade front panel MGMT OOB (4 in Figure 10.3) to your PC.
Chapter 11 Diagnosing a Failed Ethernet Switch Blade Activation Figure 11.1: Ethernet Switch Blade Activation States The Ethernet Switch Blade must transition through a series of states (M0–M4) to become active in an ATCA shelf. After the Ethernet Switch Blade has reached the M4 state, it will become active and start the boot process of the OpenArchitect Switch Management environment.
Page 151
FRU State HotSwap Healthy Solution LED Status LED Status No power. Board not inserted correctly. 1. Remove and re-insert board. 2. If board does not power-up after re-insertion, try a different slot. If board continues to fail in the new slot and the problem does not affect other boards running in the chassis, return the Ethernet Switch Blade board for repair.
FRU State HotSwap Healthy Solution LED Status LED Status switch through a console cable. If OpenArchitect is running, and abnormal behavior is occurring, please see Network Configuration Problems for information on network issues. If OpenArchitect cannot be accessed through the console port, please see Troubleshooting a Failed OpenArchitect Load.
sensor information. Examine the System Event Log (SEL) on the ShMM and determine if critical sensor events have been logged for the switch in question. If the switch has reported critical sensor data for temperature or voltage, the ShMM can prevent it from booting.
Page 154
clia board -v 7 clia board -v 8 These commands generate an output that reports if the ShMM thinks it has granted access to ports on the switches. Check the Shelf Manager User’s Guide for the expected output. Ethernet Switch Blade User's Guide release 3.2.2j page 154 Downloaded from...
Chapter 12 Troubleshooting a Failed OpenArchitect Load The OpenArchitect operating system is loaded from the FlashROM memory into RAM when the Ethernet Switch Blade is activated by the Shelf Manager. If there is a problem with the loading of OpenArchitect due to a hardware failure or corrupt file system, the back-up image can help to troubleshoot the condition.
Ethernet Switchblade has Ethernet Switchblade has been enabled by the been enabled by the ShMM and starts to boot ShMM and starts to boot Bootloader examines the bootstring in the Boot Loads image from Flash Determines if device 1 the bootstring is dev 1 Loads image from Flash Determines if...
If you cannot telnet, attach a console cable and Modem Eliminator Cable to the switch. A console cable ( PN A6900-63006) is included with each HP bh5700 ATCA 14-Slot Blade Server, and a Modem Eliminator Cable should be obtainable locally. You can also obtain the correct console cable from your Hewlett-Packard sales representative.
properly attach the console cable. Booting Without the Overlay File If you cannot telnet into the switch and Linux fails to boot, it is likely that a change saved by zsync has left the switch in an inaccessible state. To allow users to recover from mistakes saved in the overlay file system, a boot argument of –i passed to the init process will stop the untarring of the saved overlay files.
If the switch still is unable to boot, see Booting the Duplicate Flash Image, below. Booting the Duplicate Flash Image Another recovery method, if Linux fails to boot, is to temporarily boot the factory-installed duplicate image located in the second flash device. 1.
Chapter 13 Network Configuration Problems Many reported problems on a booted switch will ultimately be traced back to user errors in the layer 2 or layer 3 switch configuration. In some cases, symptoms from an improperly configured switch can masquerade as potential hardware problems. Interface Overview On startup OpenArchitect creates interfaces for all Ethernet ports on the Ethernet Switch Blade.
Physical Slot Fabric Port 3 Fabric 51** * Base Interface Inter-Switch Link (ISL) ** 10 Gigabit Ethernet Fabric Interface - Update Channel Table 13.2: Additional Interfaces Additional Interfaces Base Fabric Front Panel Egress (zre) Shelf Manager 1 (zre) Shelf Manager 2 (zre) Front Panel Out-of-Band Management Port (eth) NOTE: The Out-of-Band (eth) ports are not enabled by default.
Page 162
2. S30e1000 Script that loads the e1000 driver module for the Out-of-Band Ethernet ports. (Editing this script is not recommended.) Script that checks the current OA version, and loads into the Vital Product S40vpd Data (VPD) area if necessary. (Editing this script is not recommended.) 3.
Configuration Troubleshooting Problem Solution No Connection Physical Link problem. Check to see if the port LED is lit. If the LED port is not lit, then you may have a bad cable connection. Configuration Error. Connect through the console port (See Chapter 10).
Page 166
The following table will translate the zlc output to link status. Link Port Status Link Speed Pause Faults EKEY_DISA Auto Enable Internal Fault BLED EKEY_ENAB 1000fd Disable External Fault 1000hd DOWN 100fd 100hd 10fd 10hd Link: zre(X) – physical interface Shelf Manager Status: EKEY_DISABLED - A slot or device that has been disabled by the Shelf Manager.
10hd – Ethernet Half Duplex Pause: Enable: a port that can temporarily suspend the data transmission between two network devices in the event that one of the devices becomes congested. Pause enabled devices can reduce bottlenecks by making the network more efficient. Disabled: The pause feature is not enabled and will continue to transmit traffic when even when the receiving device is busy.
sh-2.04# zlc zre0..23 query zre0: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre1: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre2: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre3: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre4: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON>...
Page 169
[ZX7100-OA3.2.2h]# zlc zre0..51 query zre0: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre1: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre2: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre3: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre4: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON>...
Network Connectivity Troubleshooting No Connection If the port LED is lit on the front panel, the switch has established a physical connection and the problem is a network configuration error. Check to see if both devices are configured to be on the same network (ex.
have an active remote device attached, then first bring down the ports which do not have active connections expected to make sure there is a legitimate EXT FLT condition. If loss of communications is suspected on an externally wired port, make sure to check and test affected cables.
Traceroute Test It’s possible to trace a network path using the traceroute command. The following is an example of a Layer 2 traceroute with only two devices. sh-2.04# traceroute 192.168.1.101 traceroute to 192.168.1.101 (192.168.1.101), 64 hops max, 40 byte packets 1 192.168.1.101 (192.168.1.101) 1.888 ms 1.135 ms 0.814 ms sh-2.04# Ethernet Switch Blade User's Guide release 3.2.2j...
6. Isolation Transformer 13. Isolation Transformers 7. Power Supply 14. 4 Port PHY Hardware Subsystem In the following tables, refer to the identified component-area numbers on indicated in the pictures in the proceeding section. The indications of malfunction may be identified either during normal operation, or in response to a specific test.
Base ZMC 0 ZMC 1 Hardware Indications of Malfunction Subsystem any of the following indications: Error message via OpenArchitect due • to inability to access the registers within the switch chip, or a failure of DMA transfers. Loss of switch functionality, such as •...
Duplicate Flash Image. If the switch can successfully boot from FlashROM device 2, then FlashROM device 2 is fully operational. Testing the Switch Fabric You can test the functionality of the switch fabric by running the zlc command. The zlc command outputs the link status for any Ethernet Switch Blade interface.
Example Output: sh-2.04# zlc zre0..23 query zre0: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre1: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre2: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre3: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON> zre4: <EKEY_DISABLED, AUTO, PAUSE ENABLE, EXT_FLT ON, OK ON>...
If the “Used” and “Free” memory statistics do not add up to the Total memory, the software environment may have a memory leak caused by a software error. Reboot the switch. If the problem persists after a reboot. Run the top command to list the memory utilization of all current processes.
To test the operational status of the control processors you can do the following: Hardware Fault Connect to the console port of either the Base or Fabric Interface control processor (See Chapter 10 for more information). If you cannot communicate with the Ethernet Switch Blade, the control processor may have encountered a software error.
Page 182
INT FLT LED is illuminated, replace the switch and return it for repair. Ethernet Switch Blade User's Guide release 3.2.2j page 182 Downloaded from www.Manualslib.com manuals search engine...
Chapter 15 High Availability Troubleshooting The ATCA environment will usually contain a high-availability failover configuration between two ATCA switches in the chassis. Note that the failover features are configurable and a switch can be directed to fail over all of its processing when a single port or link goes down, or it can perform a port-to-port or VLAN-to-VLAN failover where both partner switches are still processing a portion of the network traffic.
3. IPMI firmware Some hardware and software problems can be resolved by updating the firmware to the latest version. Check the Hewlett-Packard website for the latest version (see the HP 5700 ATCA 14- Slot Blade Server Installation Guide). Checking the switch firmware version Use the zstats –V command to output the Vital Product Data from switch memory.
Key: PN: Base Interface Switch Assembly Number SN: Base Interface Switch Serial Number V6: OpenArchitect Version Number VP: IPMI Firmware Version VZ: BootLoader Version Number The following output is shown for the 3.1 Fabric Interface: 3.1 Fabric Interface [ZX7100-OA3.2.2h]# zstats -V VITAL PRODUCT DATA: Open Architect Advanced TCA Fabric Switch PN = 700-0174-002...
-d 0 <bootloader image name> OpenArchitect Firmware Upgrade: 1. Refer to the HP bh5700 14-Slot Blade Server Installation Guide, Chapter 6, 14-Slot Shelf Startup, Validating and Updating Your Firmware for instructions on how to gain access to firmware updates for the HP bh5700 14-Slot Blade Server.
Surviving Partner daemons to think there is a failure, resulting in link oscillation. Base Interface: zflash -d 1 rdr6000.zImage.initrd Fabric Interface: zflash –d 1 rdr7100.zImage.initrd IPMC Firmware Upgrade: Upgrading the IPMC Firmware through OpenArchitect is not currently supported. Ethernet Switch Blade User's Guide release 3.2.2j page 187 Downloaded from...
Chapter 17 Restoring the Factory Default Configuration You should use this procedure if the contents in Flash Device 1 are corrupt and you need to restore the switch to the factory default configuration. By restoring the factory default configuration, you will overwrite your main file system in Flash Device 1 and lose all previous configuration changes.
Chapter 18 Before Calling Support Because of the highly customized configurations that can be applied by customers to their ATCA switch environment, the focus must be on data collection to get a snapshot of the current switch configuration and network traffic activity. If support is needed, it is necessary to gather the following information for further diagnosis before calling support: 1.
Application Application Boot ROM Flash 2 on Flash 1 on on Device 0 Device 2 Device 1 Offset 0 Offset 0 initrd zmon initrd (exact copy as in Linux and Application Free space its file Flash 1) system Linux and its file system Offset 7f000 Free space...
Appendix A Fabric Switch Command Man Pages OpenArchitect applications are implemented above the OpenArchitect libraries and the RMAPI interface. OpenArchitect applications are used for normal operation of the switch, for runtime status and diagnostics, and for prototyping new applications development. For runtime operation, the OpenArchitect applications perform initialization and configuration, and real-time control and maintenance of the switching tables in the switch silicon.
vrrpconfig NAME vrrpconfig – Configure and control the running vrrpd SYNOPSIS vrrpconfig [-d <level>] -- <vrrpd parameters> vrrpconfig [-d <level>] [-k] [-a] [-p] [-s <vid>] DESCRIPTION vrrpconfig provides communication with a running vrrpd daemon. The -- option for vrrpconfig will pass all parameters to vrrpd as would be done when starting the vrrpd. Any output generated by vrrpd is displayed on the vrrpconfig controlling tty.
EXAMPLES Here is an example of using the -- invocation method that changes the priority to 99 for the Virtual Router associated with the Virtual Router Identifier 1: vrrpconfig -- -v 1 –p 99 SEE ALSO vrrpd Ethernet Switch Blade User's Guide release 3.2.2j page 193 Downloaded from...
vrrpd NAME vrrpd – Virtual Router Redundancy Protocol Daemon SYNOPSIS vrrpd -i ifname -v vrid [-f piddir] [-s] [-a auth] [-p prio] [-nhb] [-I ifname] [-d delay] [-m address] [-M ] [-B] [-S script] [-c conf_file] [-D level] ipaddr DESCRIPTION vrrpd is an implementation of Virtual Redundant Routing Protocol (VRRPv2) as specified in RFC2338.
Page 195
the –i option. Toggle preemption mode (Enabled by default). Preemption means that a Master switch will go to Backup if a current Backup has higher priority. Become MASTER when priority is equal. Be sure it is only set on one host or the switches will oscillate. Must set –B option on other hosts (requires preemption mode ! -s) Become BACKUP when priority is equal.
Page 196
The result of which to use the native MAC address of the interface. Using the –n option is not recommended. -D <level> Set debugging output to the supplied level <ipaddr> the ip address(es) of the virtual server SEE ALSO vrrpconfig Ethernet Switch Blade User's Guide release 3.2.2j page 196...
zbootcfg NAME zbootcfg − Modifies the boot parameters of the OpenArchitect switch. SYNOPSIS zbootcfg -a | -d <device number> [<boot_string>] DESCRIPTION zbootcfg is used to display or modify the boot parameters on the switch. The boot parameters are utilized by the minof boot loader application to indicate on which device to find a boot image. Care should be taken when changing the boot string.
Page 198
application flash. Typically this is required before updating application flash 1. By booting the alternative image, if a failure occurs during the programming of application flash 1, recovery is easier. zbootcfg -d 2 The next example passes the -i option to the booting kernel. This is useful when recovering from a mistake saved to the read-write file system or after updating the application flash 1 and doing the first boot.
zconfig NAME zconfig - Configures the OpenArchitect switch. SYNOPSIS zconfig [-h <host_name>] [-d <level>] [-a] [-t] [{-f <file>} | <configuration>] DESCRIPTION zconfig creates Virtual Local Area Network (VLAN) groups of switch ports or trunks. Each VLAN group forms a Layer 2 switching domain. Each VLAN group has a VLAN Identification number (VID) that can be carried in a tag field, located in the header of packets traveling on that VLAN.
input. If the -f flag is not used, a single line of configuration data can be entered as parameters to zconfig. CONFIGURATION SYNTAX zconfig takes configuration data from standard input or from a file with the -f option. In either case, the configuration syntax is the same. The zconfig configuration data consists of a list of semicolon-delimited statements.
Page 201
ports. teardown Removes the trunk interface, making the ports which were part of the trunk available for configuration in other trunks or VLANs. mac [ source_address | destination_address ] ip [ source_address | destination_address ] port [ source_port | destination_port ] Further specifies the rules for selecting which port in the trunk a packet should be transmitted out of.
Page 202
The syntax for a network interface statement is the interface name followed by a colon and an action. Network interface statements are used to create or tear down a VLAN group and can consist of one or a list of network interface names; followed by a colon and then an action. For example: zhp0: <Network Interface Action>;...
Page 203
named zhp1. This VLAN includes four switch ports, zre1, zre10, zre11, zre13. zhp0: vlan100 = zre1,zre10,zre11,zre13; The next statement adds two switch ports, zre1, zre2 and zre3, to an existing network interface and VLAN. zhp0: vlan100 = zre1..3; The next statement deletes two switch ports, zre1 and zre2, to an existing network interface and VLAN.
Page 204
zre1=untag1; If port 0 is also a member of VLAN 100, packets for VLAN 100 are sent from this port with a VLAN tag as part of their header. In the next example, the switch ports 10, 11, and trunk 2 are configured as untagged members of VLAN 100.
Page 205
The following statement creates 14 VLAN groups with VID numbers 1-14. Each VLAN contains the same switch port, port 1, represented as zre1. zhp0..13: vlan1..14 = zre1; The plus (+) wildcard can be used with the last port listed to auto-increment that port number before each VLAN group is created.
Page 206
The previous configuration can be used for creating a 14 port Layer 2 switch, all 14 ports assigned to the same VLAN. SEE ALSO zl3d Ethernet Switch Blade User's Guide release 3.2.2j page 206 Downloaded from www.Manualslib.com manuals search engine...
zcos NAME zcos - class of service queue control SYNOPSIS zcos [-h <hostname>] [-d <level>] [ -u <default priority> ] [ -m q0,q1,q2,q3,q4,q5,q6,q7 ] [-n <queue length list in packets for each queue> | -b <Reserved space in bytes for each queue> | -s <limit on dynamic pool usage, in bytes>, <reset %>] [ -k PRI | RR | WRR | DRR] [ -w <queue weight list>...
Page 208
Each cos queue is limited in the number of packets it can hold waiting scheduling; the memory used by each queue is managed to provide a guaranteed space with additional space shared among all queues for a port. OPTIONS Most options are optionally followed by a <port list>, which may include zre port ranges, like zre0..5, individual ports, such as zre51, or cpu, to indicate the queues and scheduling for packets to be transferred to the cpu.
Page 209
Packets are first counted against the reserved space for a queue. When that space is occupied, additional memory is used from the dynamic memory pool until the dynamic pool usage limit for the port is reached. Any additional packets received for the queue on this port are dropped. Metering and Scheduling -r <list of bandwidth guarantees in Kbps for each cos queue>...
Page 210
-q all | qmap | qinfo | scheduler [<port list>] Queries the current COS/QOS Settings. all - Displays all of the queue mappings, queue limits, metering and scheduling settings qmap - Displays the priority to COS queue mappings. qinfo - Displays queue limits for the COS queues. scheduler - Displays the traffic metering and shaping settings and the scheduler mode.
zdog NAME zdog - Configure and send heartbeats to watch dog enabled drivers. SYNOPSIS zdog [-d <level>] -h | -i <interval> | -n <heartbeats> zdog [-d <level>] -b zdog [-d <level>] -a DESCRIPTION zdog is used to configure the Ethernet Switch Blade watchdog timer functions and to send heartbeats to the Ethernet Switch Blade watchdog drivers.
Page 212
component of zdog insures that the CPU is functioning well enough to execute something. The software component of zdog when launched from monit insures that monit is running to perform higher level tasks. And finally monit can be used to monitor any or all critical system resources and processes in the system.
zfilterd NAME zfilterd - A daemon to use the filter hardware of the OpenArchitect switch for filtering based on iptables(8) rules. SYNOPSIS zfilterd [-d <level>] [-p <port>] [-f] [-l] [-i <pid>] [-o <pid>] DESCRIPTION zfilterd is a daemon that intercepts filtering rules entered by the user, using iptables(8), checks them for validity and then prepares messages for the traffic management daemon ztmd, which is responsible for setting up the switch hardware for the filtering rules and actions.
zflash NAME zflash − Loads images into the flash ROMs on the OpenArchitect switch. SYNOPSIS zflash -d <dev> [-o|-O <offset>] <image_file> <upgradeipmi.img> DESCRIPTION zflash enables you to program the flash ROMs on the switch. The switch contains 3 flash ROM devices: the boot ROM flash, application flash 1 and application flash 2. Care should be taken when flashing new images into the switch.
Page 215
Exercise caution when using this command, as an error can render your switch inoperable. Do not interrupt this process until complete. SEE ALSO zbootcfg Ethernet Switch Blade User's Guide release 3.2.2j page 215 Downloaded from www.Manualslib.com manuals search engine...
zl2, zl2mc, zl3host, zl3net, zvlan NAME zl2, zl2mc, zl3host, zl3mc, zl3net, zvlan – Formatted display of OpenArchitect generic tables. zl2 displays the abstraction API’s layer 2 table. zl2mc displays the abstraction API’s layer 2 multicast table. zl3host displays the abstraction API’s layer 3 host route table. zl3mc displays the abstraction API’s layer 3 multicast table.
Page 217
display on the user console. The format of the output is table-dependent. Port mapping affects the ports referenced in the generic tables. (Ports listed in order from 1 to 15) Headers describing the column being displayed are printed after every 22 lines of output, which makes it easy to pipe through more(1).
Page 218
The following command deletes the above entry: zl2 -c -m 00:c0:95:45:00:00 The following command displays all entries of the zl2 table: Be careful, the -c option does not ask. The following command deletes all entries in the zl2 table: zl2 -c SEE ALSO Ethernet Switch Blade User's Guide release 3.2.2j...
zgvrpd NAME zgvrpd - GARP VLAN Registration Protocol (GVRP) daemon for the OpenArchitect switch. SYNOPSIS zgvrpd [-d <level>] [-f] [-h <hostname>] [-p <ppa>] [-t <target>] DESCRIPTION zgvrpd is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).zgvrpd starts a background task that implements the GARP VLAN Registration Protocol (GVRP) protocol for a specified zhp interface.
Page 220
OPTIONS -d <level> Sets the level of debugging output required by zgvrpd. The default level is zero (0). Setting the debug level higher produces more output. Five (5) is currently the maximum output level. Run zgvrpd in foreground. Default is to run it in background.
zl2d NAME zl2d - Layer 2 daemon for the OpenArchitect switch. SYNOPSIS zl2d [start | stop] [-t <msecs>] [-d <level>] [-f] [-p <priority>] <iface..> DESCRIPTION zl2d is run after the network interfaces are created and initialized with zconfig. zl2d creates a Linux bridge for each interface using brctl(8). The bridge name is the interface name with a ‘b’...
Page 222
OPERATIONS zl2d manages the Spanning Tree state fields in the switch of each port within the bridge(s). Based on a timer, zl2d reads the port information for each Linux bridge and updates the switch when necessary. EXAMPLES In the following example, zl2d creates a Linux bridge named bzhp0 which includes all of the zre<n>...
zl3d NAME zl3d - Layer 3 daemon for the OpenArchitect switch. SYNOPSIS zl3d [-h <host_name>] [-t <msecs>] [-b] [-e] [-l] [-n] [-d <level>] <iface ..> DESCRIPTION zl3d is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).zl3d listens for Netlink messages from the kernel and monitors the Linux network routing tables for routing updates.
Page 224
produces more output. Four (4) is currently the maximum output level. <iface…> The network interfaces on which zl3d should operate. These network interfaces must first be created by zconfig. zl3d does not operate with standard network interface cards. It only works on switch network interfaces created by zconfig.
NAME zlc − link and LED control SYNOPSIS zlc [-h <hostname>][-d <level>][-x] <port_list> <action> [on | off ] zlc [-h <hostname>][-d <level>][-x] <action> [on | off |clear] zlc [-h <hostname>][-d <level>][-x] [state|query] DESCRIPTION The zlc application sets the link speed and state of individual ports of the switch, or displays the current state.
Page 226
EXAMPLES In the following example, zlc forces the line speed of port 1 to 100 Full duplex. The interface must be down to change the speed. Assuming zre1 is part of interface zhp0, ifconfig zhp0 down zlc zre1 100fd The external fault, internal fault, and ok LEDs can be set on a per port basis or globally. To set the external fault LED for a particular port, zlc zre1 extfault on To query the settings of a particular port,...
Page 227
SEE ALSO ifconfig(8) Ethernet Switch Blade User's Guide release 3.2.2j page 227 Downloaded from www.Manualslib.com manuals search engine...
zlmd NAME zlmd − monitor link changes or hot swap events. SYNOPSIS zlmd [-h <hostname>] [-b] [-d <level>] {-f <file>} | <configuration> DESCRIPTION The zlmd application is intended to run as a daemon, waiting for a configured event to occur and then running the program configured for that event.
Page 229
<port-list> A list of ports in the same forms supported by zconfig, e.g. zre1,zre2 or zre10..14 <program> Path to an executable program or script to be run when the event occurs. Note: An absolute path to <program> is required. The program will be called with the following parameters: For Link Changes: <program>...
zlogrotate NAME zlogrotate − Rotates log files. SYNOPSIS zlogrotate [-b] [-t time] [-s segment size] [-n # of files] [-f file to rotate] DESCRIPTION zlogrotate rotates the selected file every [time] seconds if the file is larger than [segment size]. It will keep only the number of files selected. zlogrotate is called from /etc/init.d/rcS by default with no parameters.
zmirror NAME zmirror - Set packet mirroring on an ingress or egress port. SYNOPSIS zmirror -a | -t zmirror [-e] <from_list> <to_port> DESCRIPTION zmirror sets packet mirroring from a given set of ports to a given port. Turning on packet mirroring causes a copy of the packet to be sent to the to port.
Page 232
zmirror is cumulative: zmirror zre1, zre2, zre3 cpu Is the same as: zmirror zre1 cpu zmirror zre2 cpu zmirror zre3 cpu Setting a different to port will overwrite the previous setting and direct previously mirrored ports to a new to port. Given the last setup the following will change port 1 traffic to be forwarded to port 10.
zmnt NAME zmnt − Expands the read/write files onto the RAM disk. SYNOPSIS zmnt [-c] <directory> zmnt [-c] -t <file> zmnt [-c] –l DESCRIPTION zmnt expands files from flash onto the RAM disk that have been previously saved with zsync. The init process runs zmnt to expand the files in flash onto RAM file system.
Page 234
In the following example, zmnt the current overlay into a tar file called overlay.tar zmnt –t overlay.tar The resulting tar file can now be saved on a different host as a snapshot of the overlay at that point in time. Use zsync to restore the overlay on the switch: zsync –t overlay.tar The restored overlay will be used upon the next reboot.
zpeer NAME zpeer – Application for High Availability communication between the Fabric and Data switches. SYNOPSIS zpeer [-d <level>] local|peer <command> <value>|query zpeer [-d <level>][-a][-r] DESCRIPTION zpeer is used to pass bidirectional High Availability(HA) state and priority information between the base and fabric switches in the Ethernet Switch Blade. zpeer uses the concept of local and peer information.
Page 236
displayed by query, but must be set at initialization. After setting the healthy state, the query will return the backup state. backup Used to reflect the backup state of vrrpd master Used to reflect the master state of vrrpd The priority value is a value between 0 and 255. In the HA suite, the value is set to 254 minus the number of ports that are link down.
Page 237
be also reset. SEE ALSO zspconfig Ethernet Switch Blade User's Guide release 3.2.2j page 237 Downloaded from www.Manualslib.com manuals search engine...
zqosd NAME zqosd – monitors tc(8) commands to implement classification filters and queuing disciplines in hardware. SYNOPSIS zqosd [-d <level>] [-p <port>] [-f] [-l] [-i <pid>] [-o <pid>] DESCRIPTION zqosd monitors commands entered by tc which set up queuing disciplines and classification filters for managing traffic in the switch.
Page 239
SEE ALSO ztmd, tc(8), zfilterd Ethernet Switch Blade User's Guide release 3.2.2j page 239 Downloaded from www.Manualslib.com manuals search engine...
NAME zrc - Packet rate control SYNOPSIS -b | -m | -d | -t | -a [-p <port>] [-v <vlan>] [-g <group>] [-M <mac_addr>] [-T <timeout>] [-D <level>] <rate> DESCRIPTION zrc sets rate control on Broadcast, Multicast and/or Destination Lookup Failure (DLF) packets. The rate is measured in the number of packets per time period.
zreg NAME zreg - Read and write registers and tables on the OpenArchitect switch switching hardware. SYNOPSIS zreg [-p <ppa>] [-w] [-i <index>] [-t <index>] [-k] [-h <hostname>] [-d <level>] [-r 10] <reg> DESCRIPTION zreg allows a user to read and write direct and indirect registers and tables on the resident switch chip.
Page 242
default zreg configures the OpenArchitect switch that is locally connected (i.e., the one that is on the local PCI bus). -r 10 Sets numeric radix for registers to 10. Default is 16. -d <level> Set the level of debugging output produced by zreg. The default level is 1.
zrld NAME zrld – ZNYX redirector daemon SYNOPSIS zrld [-d <level>] [-p <port>] [-f] DESCRIPTION zrld is used for remote management of OA/HA applications. OA/HA applications capable of remote management include zlc, ztats, zlmd. zrld only allows requests from hosts listed in /etc/rcZ.d/zrld_trusted_hosts. OPTIONS -d <level>...
zsnoopd NAME zsnoopd - IGMP Snooping daemon for the OpenArchitect switch. SYNOPSIS zsnoopd [-d <level>] [-f] [-h <hostname>] [-p <ppa>] [-r <sec>] [-t <sec>] [-u <sec>] [-v <vlan_id>] DESCRIPTION zsnoopd is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).zsnoopd starts a background task that monitors incoming IGMP traffic in order to learn which hosts in a VLAN are listening to which IP multicast addresses.
Page 245
Run zsnoopd in foreground. Default is to run it in background. -h <hostname> Connect to remote host <hostname>. -p <ppa> Start zsnoopd on switch <ppa>. Default is 0. -r <sec> Time to wait, in seconds, before removing a port with no router multicast traffic. Default is 260 seconds. -t <sec>...
zspconfig NAME zspconfig - configure and start surviving partner SYNOPSIS zspconfig [-d <level>] [-p <directory_path>] [-u <dhcp_interface>] [-c <dhclient.conf>] [-t <timeout>] [-s] [-v] -f <file> DESCRIPTION zspconfig is used to configure and start the Surviving Partner software. With the -f option a configuration file is provided that completely describes the network setup and desired behavior of all of the switches participating in the Surviving Partner.
-t <timeout> Time to wait in seconds before giving up on finding a Surviving Partner to retrieve configuration information from. Only valid with the –u option. Do not ask for confirmation. Run from a script. Prints the current version of zspconfig. -f <file>...
Page 248
A sibling_addresses statement is required for each VLAN created with the zconfig commands. The two addresses in the list indicate there are two switches in the Surviving Partner group. The first address 10.0.0.30 and 11.0.0.30 are assigned to the switch on which the configuration is being run.
Page 249
monitor_only - Ports that are monitored but do not have a virtual address managed on them. They will not have their links brought down temporarily during a failover scenario. These ports are only monitored. If a problem occurs on this type of link it will cause a failover scenario.
Page 250
interconnect, it is not a bonding driver enabled port, and therefore should be setup as an interconnect port type. To accomplish this, the zre ports are listed to avoid conflicting port types. Note that a single line cannot contain both zhp and zre definitions. Therefore rain link: zhp1, zre1..4 does not work and the definition zre1..8 is equivalent.
Page 251
Additional startup scripts may be included in the configuration using the start_script command. The files in the start_script command will be placed in a location for tftp transfer to sibling switches that initialize using the –u option. A common use of the start_script command might be to propagate gated configurations to all members of the Surviving Partner group.
Page 252
distinguish potentially more than one backup switch. This configuration file is placed in /tftpboot, and is retrieved via DHCP by a replacement switch on boot up. /etc/rcZ.d/surviving_partner/dhcpd.conf Configuration script used by dhcpd when the switch becomes master. dhcpd is used to serve replacement switches their configuration scripts.
zstack NAME zstack - Configures the OpenArchitect switch stacking. SYNOPSIS zstack [-h <host_name>] [-d <level>] [-a] [-t] [{-f <file>} | <configuration>] DESCRIPTION zstack combines multiple switch fabric chips into a single virtual switch. zstack must be run before any other switch configuration. Specifically it must be run before zconfig. zstack is typically run from an S20stack script prior to the S50xxx scripts.
Page 254
semicolon-delimited statements. Each statement specifies an action to take on a stack. A stack is a group of ports on a single switch fabric chip. Actions include stack creation, stack port association, stack configuration and stack control. Comments, spaces and new lines are ignored. Comments begin with the # character and include characters through the next new line.
Page 255
an equal number of ports in each list. Wild cards may be used in the zre_lists. See below. Stack port association syntax for a Ethernet Switch Blade: stack0: zre0..11 = zre0..11; stack1: zre12..23 = zre0..11; The first statement above configures the first switch silicon chip, represented by stack0, to have no translation between its physical port numbering and its virtual port numbering.
Page 256
are supported. enable; The enable statement turns on stacking that has been previously configured. This statement cannot be made until configuration is complete. disable; The disable statement turns off stacking. Before disabling stacking, all Ethernet Switch Blade daemons must be stopped, and the VLAN configurations must be torn down using zconfig. EXAMPLES zstack stack0: ppa0 local zstack stack1: ppa1 local...
Page 257
zre lists. Example of stack0..3 representing stacks 0, 1, 2 and 3. SEE ALSO zconfig Ethernet Switch Blade User's Guide release 3.2.2j page 257 Downloaded from www.Manualslib.com manuals search engine...
ztats NAME ztats − Display statistics and information about switch SYNOPSIS ztats [-d <level>] [-i <unit>] | [-m <port>] | [-v <vlan id>] | [-t <tgid>] | [-v] DESCRIPTION ztats displays MIB counters for a selected physical port, trunk group or VLAN. It can also display information about the configuration of the switch and bridge to the PCI bus or the Vital Product Data memory.
zsync NAME zsync − Saves changes to the flash. SYNOPSIS zsync [-c][-f][<dir_or_file>] zsync [-c][-f][-t <file>] zsync [-c][-f][-z] zsync [-c][-l] DESCRIPTION zsync is used to save a snapshot of the current file system to flash ROM. By default, zsync creates a compressed tar image of the files that have changed and saves the image in the flash ROM.
Page 260
-t <file> Read files to be saved from a tar file. Zero the overlay area. Do not confirm with user and do not warn if saving failed. Exit code can be examined to determine success or failure. <dir_or_file> Save only the named file, or save the named directory to the overlay.
ztmd NAME ztmd – traffic management daemon which accepts messages from traffic filtering and quality of service applications and sets up hardware. SYNOPSIS ztmd [-d <level>] [-p <port>] [-f] [-i <pid>] [-o <pid>] [-a <addr>] [-l] DESCRIPTION ztmd listens for messages on a multicast port. These messages describe packet filters and queuing disciplines that are to be installed in the switch hardware.
Page 262
SEE ALSO zqosd, iptables(8), tc(8), zfilterd Ethernet Switch Blade User's Guide release 3.2.2j page 262 Downloaded from www.Manualslib.com manuals search engine...
brctl(8) NAME brctl - Bridge and Spanning Tree Protocol administration. SYNOPSIS brctl [options] DESCRIPTION brctl is used to set up, maintain, and display the bridge configuration in the Linux kernel. brctl is a standard command included with Linux bridge support which includes Rapid Spanning Tree Protocol (RSTP) support.
Page 264
bridge will not send or receive BPDUs, and will thus not participate in the Spanning Tree Protocol. If your bridge isn't the only bridge on the LAN, or if there are loops in the LAN's topology, DO NOT turn this option off. Turning this option off may impair network traffic, so be careful.
Page 265
SEE ALSO zconfig , zl2d Ethernet Switch Blade User's Guide release 3.2.2j page 265 Downloaded from www.Manualslib.com manuals search engine...
Appendix B Base Switch Command Man Pages OpenArchitect applications are implemented above the OpenArchitect libraries and the RMAPI interface. OpenArchitect applications are used for normal operation of the switch, for runtime status and diagnostics, and for prototyping new applications development. For runtime operation, the OpenArchitect applications perform initialization and configuration, and real-time control and maintenance of the switching tables in the switch silicon.
vrrpconfig NAME vrrpconfig – Configure and control the running vrrpd SYNOPSIS vrrpconfig [-d <level>] -- <vrrpd parameters> vrrpconfig [-d <level>] [-k] [-a] [-p] [-s <vid>] DESCRIPTION vrrpconfig provides communication with a running vrrpd daemon. The -- option for vrrpconfig will pass all parameters to vrrpd as would be done when starting the vrrpd. Any output generated by vrrpd is displayed on the vrrpconfig controlling tty.
Page 268
EXAMPLES Here is an example of using the -- invocation method that changes the priority to 99 for the Virtual Router associated with the Virtual Router Identifier 1: vrrpconfig -- -v 1 –p 99 SEE ALSO vrrpd Ethernet Switch Blade User's Guide release 3.2.2j page 268 Downloaded from...
vrrpd NAME vrrpd – Virtual Router Redundancy Protocol Daemon SYNOPSIS vrrpd -i ifname -v vrid [-f piddir] [-s] [-a auth] [-p prio] [-nhb] [-I ifname] [-d delay] [-m address] [-M ] [-B] [-S script] [-c conf_file] [-D level] ipaddr DESCRIPTION vrrpd is an implementation of Virtual Redundant Routing Protocol (VRRPv2) as specified in RFC2338.
Page 270
other words there is a unique vrid to ifname associated with the –i option. Toggle preemption mode (Enabled by default). Preemption means that a Master switch will go to Backup if a current Backup has higher priority. Become MASTER when priority is equal. Be sure it is only set on one host or the switches will oscillate.
Page 271
spaces. The –n option overrides the change made with –m. result of which to use the native MAC address of the interface. Using the –n option is not recommended. -D <level> Set debugging output to the supplied level <ipaddr> the ip address(es) of the virtual server SEE ALSO vrrpconfig Ethernet Switch Blade User's Guide...
zbootcfg NAME zbootcfg − Modifies the boot parameters of the OpenArchitect switch. SYNOPSIS zbootcfg -a | -d <device number> [<boot_string>] DESCRIPTION zbootcfg is used to display or modify the boot parameters on the switch. The boot parameters are utilized by the minof boot loader application to indicate on which device to find a boot image. Care should be taken when changing the boot string.
Page 273
application flash. Typically this is required before updating application flash 1. By booting the alternative image, if a failure occurs during the programming of application flash 1, recovery is easier. zbootcfg -d 2 The next example passes the -i option to the booting kernel. This is useful when recovering from a mistake saved to the read-write file system or after updating the application flash 1 and doing the first boot.
zconfig NAME zconfig - Configures the OpenArchitect switch. SYNOPSIS zconfig [-h <host_name>] [-d <level>] [-a] [-t] [{-f <file>} | <configuration>] DESCRIPTION zconfig creates VLAN groups of switch ports or trunks. Each VLAN group forms a Layer 2 switching domain. Each VLAN group has a VLAN Identification number (VID) that can be carried in a tag field, located in the header of packets traveling on that VLAN.
Page 275
entered as parameters to zconfig. CONFIGURATION SYNTAX zconfig takes configuration data from standard input or from a file with the -f option. In either case, the configuration syntax is the same. The zconfig configuration data consists of a list of semicolon-delimited statements. Each statement specifies an action to take globally or on an interface.
Page 276
A port member is identified with the zre<X> format, where x represents a port number between 0 and 23 for the in-band ports. The Out-of-Band ports cannot be included in the List of ports. teardown Removes the trunk interface, making the ports which were part of the trunk available for configuration in other trunks or VLANs.
Page 277
Network Interface Statements The syntax for a network interface statement is the interface name followed by a colon and an action. Network interface statements are used to create or tear down a VLAN group and can consist of one or a list of network interface names; followed by a colon and then an action. For example: zhp0: <Network Interface Action>;...
Page 278
The next statement creates a VLAN group with the VID number 100 and the network interface named zhp1. This VLAN includes four switch ports, zre1, zre10, zre11, zre13. zhp0: vlan100 = zre1,zre10,zre11,zre13; The next statement adds two switch ports, zre1, zre2 and zre3, to an existing network interface and VLAN.
Page 279
tag are given the VLAN tag with the VID number 1, enter: zre1=untag1; If port 0 is also a member of VLAN 100, packets for VLAN 100 are sent from this port with a VLAN tag as part of their header. In the next example, the switch ports 10, 11, and trunk 2 are configured as untagged members of VLAN 100.
Page 280
network interface, and all ports listed after the equal sign are included in each group. The following statement creates 14 VLAN groups with VID numbers 1-14. Each VLAN contains the same switch port, port 1, represented as zre1. zhp0..13: vlan1..14 = zre1; The plus (+) wildcard can be used with the last port listed to auto-increment that port number before each VLAN group is created.
Page 281
The previous configuration can be used for creating a 14 port Layer 2 switch, all 14 ports assigned to the same VLAN. SEE ALSO zl3d Ethernet Switch Blade User's Guide release 3.2.2j page 281 Downloaded from www.Manualslib.com manuals search engine...
zcos NAME zcos - class of service queue control SYNOPSIS zcos [-h <hostname>] [-d <level>] [ -u <default priority> ] [ -m q0,q1,q2,q3,q4,q5,q6,q7 ] [-n <queue length list in packets for each queue> | -b <Reserved space in bytes for each queue> | -s <limit on dynamic pool usage, in bytes>, <reset %>] [ -k PRI | RR | WRR | DRR] [ -w <queue weight list>...
Page 283
Each cos queue is limited in the number of packets it can hold waiting scheduling; the memory used by each queue is managed to provide a guaranteed space with additional space shared among all queues for a port. OPTIONS Most options are optionally followed by a <port list>, which may include zre port ranges, like zre0..5, individual ports, such as zre51, or cpu, to indicate the queues and scheduling for packets to be transferred to the CPU.
Page 284
Packets are first counted against the reserved space for a queue. When that space is occupied, additional memory is used from the dynamic memory pool until the dynamic pool usage limit for the port is reached. Any additional packets received for the queue on this port are dropped. Metering and Scheduling -r <list of bandwidth guarantees in Kbps for each cos queue>...
Page 285
-q all | qmap | qinfo | scheduler [<port list>] Queries the current COS/QOS Settings. all - Displays all of the queue mappings, queue limits, metering and scheduling settings qmap - Displays the priority to COS queue mappings. qinfo - Displays queue limits for the COS queues. scheduler - Displays the traffic metering and shaping settings and the scheduler mode.
zdog NAME zdog - Configure and send heartbeats to watchdog enabled drivers. SYNOPSIS zdog [-d <level>] -h | -i <interval> | -n <heartbeats> zdog [-d <level>] -b zdog [-d <level>] -a DESCRIPTION zdog is used to configure the base switch watchdog timer functions and to send heartbeats to the base switch watchdog drivers.
Page 287
software component of zdog when launched from monit insures that monit is running to perform higher level tasks. And finally monit can be used to monitor any or all critical system resources and processes in the system. OPTIONS set debug level to <level> Toggle use of the hardware watchdog timer.
zffpcounter NAME zffpcounter—Query or clear one or more Fast Filter Processor (FFP) counters. SYNOPSIS zffpcounter -P <zre_port> [-p <ppa>] [-i <index>] <hostname>] [-c] [-d <level>] DESCRIPTION The switch enforces filtering rules through the FFP. Each filtering rule may specify an FFP counter, to be incremented for every packet that matches that rule.
Page 289
EXAMPLES The first example queries all FFP counter values. zffpcounter The output displays the initial state of the counters. Note that the counters are not initialized on startup, Counter 0: 59602801 Counter 1: 83360091 Counter 2: 83361262 Counter 29: 83074779 Counter 30: 81723249 Counter 31: 71007391 The next example clears all FFP counter values.
Page 290
Counter 30: 0 Counter 31: 0 iptables(8) is used to setup a rule, and associate that rule with a counter. For instance, add a rule to accept all packets from 10.0.0.11 and associate that rule with FFP Counter 1. iptables –A FORWARD –s 10.0.0.11 –j ZACTION -–accept –-counter 1 Start zfilterd to move the rule entered with iptables(8) down into the switching silicon.
Page 291
Counter 19: 0 Counter 20: 0 Counter 21: 0 SEE ALSO zirule, iptables(8) Ethernet Switch Blade User's Guide release 3.2.2j page 291 Downloaded from www.Manualslib.com manuals search engine...
zfilterd NAME zfilterd - A daemon to use the filter hardware of the OpenArchitect switch for filtering based on iptables(8) rules. SYNOPSIS zfilterd [-d <level>] [-p <port>] [-f] [-l] [-i <pid>] [-o <pid>] DESCRIPTION zfilterd is a daemon that intercepts filtering rules entered by the user, using iptables(8), checks them for validity and then prepares messages for the traffic management daemon ztmd, which is responsible for setting up the switch hardware for the filtering rules and actions.
zflash NAME zflash − Loads images into the flash ROMs on the OpenArchitect switch. SYNOPSIS zflash -d <dev> [-o|-O <offset>] <image_file> <upgradeipmi.img> DESCRIPTION zflash enables you to program the flash ROMs on the switch. The switch contains 3 flash ROM devices: the boot ROM flash, application flash 1 and application flash 2. Care should be taken when flashing new images into the switch.
Page 294
Exercise caution when using this command, as an error can render your switch inoperable. Do not interrupt this process until complete. SEE ALSO zbootcfg Ethernet Switch Blade User's Guide release 3.2.2j page 294 Downloaded from www.Manualslib.com manuals search engine...
zgmrpd NAME zgmrpd - GARP Multicast Registration Protocol (GMRP) daemon for the OpenArchitect switch. (Partially supported in this release.) SYNOPSIS zgmrpd [-d <level>] [-f] [-h <hostname>] [-p <ppa>] [-t <target>] DESCRIPTION zgmrpd is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).zgmrpd starts a background task that implements the GARP Multicast Registration Protocol (GMRP) protocol for a specified interface, either a zhp or bzhp.
Page 296
Only the GARP normal registration mode is currently supported. Multiple instances of zgmrpd may run concurrently provided the targets are unique. However, zgmrpd cannot run concurrently with zsnoopd. See zsnoopd. OPTIONS -d <level> Sets the level of debugging output required by zgmrpd. The default level is zero (0).
NAME zl2, zl2mc, zl3host, zl3net, zvlan – Formatted display of OpenArchitect generic tables. zl2 displays the abstraction API’s layer 2 table. zl2mc displays the abstraction API’s layer 2 multicast table. zl3host displays the abstraction API’s layer 3 host route table. zl3net displays the abstraction API’s layer 3 network route table.
Page 298
Several options are available which enable the user to display only selected entries. Additionally, there is an option that clears user-specified entries in the table. OPTIONS -i <index> Displays the entry at the <index> position in the table. Valid for all tables. Cannot be combined with -m, -P or -v -m <mac_address>...
Page 299
Be careful, the -c option does not ask. The following command deletes all entries in the zl2 table: zl2 -c SEE ALSO Ethernet Switch Blade User's Guide release 3.2.2j page 299 Downloaded from www.Manualslib.com manuals search engine...
zgvrpd NAME zgvrpd - GARP VLAN Registration Protocol (GVRP) daemon for the OpenArchitect switch. SYNOPSIS zgvrpd [-d <level>] [-f] [-h <hostname>] [-p <ppa>] [-t <target>] DESCRIPTION zgvrpd is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).zgvrpd starts a background task that implements the GARP VLAN Registration Protocol (GVRP) protocol for a specified zhp interface.
Page 301
default level is zero (0). Setting the debug level higher produces more output. Five (5) is currently the maximum output level. Run zgvrpd in foreground. Default is to run it in background. -h <hostname> Connect to remote host <hostname>. -p <ppa> Start zgvrpd on switch <ppa>.
zl2d NAME zl2d - Layer 2 daemon for the OpenArchitect switch. SYNOPSIS zl2d [start | stop] [-t <msecs>] [-d <level>] [-f] [-p <priority>] <iface..> DESCRIPTION zl2d is run after the network interfaces are created and initialized with zconfig. zl2d creates a Linux bridge for each interface using brctl(8). The bridge name is the interface name with a ‘b’...
Page 303
OPERATIONS zl2d manages the Spanning Tree state fields in the switch of each port within the bridge(s). Based on a timer, zl2d reads the port information for each Linux bridge and updates the switch when necessary. EXAMPLES In the following example, zl2d creates a Linux bridge named bzhp0 which includes all of the zre<n>...
zl3d NAME zl3d - Layer 3 daemon for the OpenArchitect switch. SYNOPSIS zl3d [-h <host_name>] [-t <msecs>] [-b] [-e] [-l] [-n] [-d <level>] <iface ..> DESCRIPTION zl3d is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).
Page 305
produces more output. Four (4) is currently the maximum output level. <iface…> The network interfaces on which zl3d should operate. These network interfaces must first be created by zconfig. zl3d does not operate with standard network interface cards. It only works on switch network interfaces created by zconfig.
NAME zlc − link and LED control SYNOPSIS zlc [-h <hostname>][-d <level>][-x] <port_list> <action> [on | off ] zlc [-h <hostname>][-d <level>][-x] <action> [on | off |clear] zlc [-h <hostname>][-d <level>][-x] [state|query] DESCRIPTION The zlc application sets the link speed and state of individual ports of the switch, or displays the current state.
Page 307
must be down to change the speed. Assuming zre1 is part of interface zhp0, ifconfig zhp0 down zlc zre1 100fd The external fault, internal fault, and ok LEDs can be set on a per port basis or globally. To set the external fault LED for a particular port, zlc zre1 extfault on To query the settings of a particular port,...
zlmd NAME zlmd − monitor link changes or hot swap events. SYNOPSIS zlmd [-h <hostname>] [-b] [-d <level>] {-f <file>} | <configuration> DESCRIPTION The zlmd application is intended to run as a daemon, waiting for a configured event to occur and then running the program configured for that event.
Page 309
<port-list> A list of ports in the same forms supported by zconfig, e.g. zre1,zre2 or zre10..14 <program> Path to an executable program or script to be run when the event occurs. Note: An absolute path to <program> is required. The program will be called with the following parameters: For Link Changes: <program>...
zlogrotate NAME zlogrotate − Rotates log files. SYNOPSIS zlogrotate [-b] [-t time] [-s segment size] [-n # of files] [-f file to rotate] DESCRIPTION zlogrotate rotates the selected file every [time] seconds if the file is larger than [segment size]. It will keep only the number of files selected. zlogrotate is called from /etc/init.d/rcS by default with no parameters.
zmirror NAME zmirror - Set packet mirroring on an ingress or egress port SYNOPSIS zmirror -a | -t zmirror [-e] <from_list> <to_port> DESCRIPTION zmirror sets packet mirroring from a given set of ports to one given port. Turning on packet mirroring causes a copy of the packet to be sent to the to port.
Page 312
zmirror zre1, zre2, zre3 cpu Is the same as: zmirror zre1 cpu zmirror zre2 cpu zmirror zre3 cpu Multiple mirroring setups can be made. The following example will mirror port 1 traffic to port 11 and port 2 traffic to port 12. zmirror zre1 zre10 zmirror zre2 zre11 Setting a different to port will overwrite the previous setting.
Page 313
(comma) Use for creating lists (dot-dot) Specifies an inclusive range Below are some examples for the correct usage of the comma (,) and dot-dot (..). Each line below produces the same results: zre1, zre2, zre3, zre4 zre1..4 zre1, zre2..4 zre1..2, zre3..4 SEE ALSO tcpdump(1M) Ethernet Switch Blade User's Guide...
zmnt NAME zmnt − Expands the read/write files onto the RAM disk. SYNOPSIS zmnt [-c] <directory> zmnt [-c] -t <file> zmnt [-c] –l DESCRIPTION zmnt expands files from flash onto the RAM disk that have been previously saved with zsync. The init process runs zmnt to expand the files in flash onto RAM file system.
Page 315
zmnt –t overlay.tar The resulting tar file can now be saved on a different host as a snapshot of the overlay at that point in time. Use zsync to restore the overlay on the switch: zsync –t overlay.tar The restored overlay will be used upon the next reboot. SEE ALSO zsync Ethernet Switch Blade User's Guide...
zpeer NAME zpeer – Application for High Availability communication between the Fabric and Data switches. SYNOPSIS zpeer [-d <level>] local|peer <command> <value>|query zpeer [-d <level>][-a][-r] DESCRIPTION zpeer is used to pass bidirectional High Availability(HA) state and priority information between the base and fabric switches in the Ethernet Switch Blade. zpeer uses the concept of local and peer information.
Page 317
displayed by query, but must be set at initialization. After setting the healthy state, the query will return the backup state. backup Used to reflect the backup state of vrrpd master Used to reflect the master state of vrrpd The priority value is a value between 0 and 255. In the HA suite, the value is set to 254 minus the number of ports that are link down.
Page 318
be also reset. SEE ALSO zspconfig Ethernet Switch Blade User's Guide release 3.2.2j page 318 Downloaded from www.Manualslib.com manuals search engine...
zqosd NAME zqosd – monitors tc(8) commands to implement classification filters and queuing disciplines in hardware. SYNOPSIS zqosd [-d <level>] [-p <port>] [-f] [-l] [-i <pid>] [-o <pid>] DESCRIPTION zqosd monitors commands entered by tc which set up queuing disciplines and classification filters for managing traffic in the switch.
Page 320
SEE ALSO ztmd, tc(8), zfilterd Ethernet Switch Blade User's Guide release 3.2.2j page 320 Downloaded from www.Manualslib.com manuals search engine...
NAME zrc - Packet rate control SYNOPSIS -b | -m | -d | -t | -a [-p <port>] [-v <vlan>] [-g <group>] [-M <mac_addr>] [-T <timeout>] [-D <level>] <rate> DESCRIPTION zrc sets rate control on Broadcast, Multicast and/or Destination Lookup Failure (DLF) packets. The rate is measured in the number of packets per time period.
zreg NAME zreg - Read and write registers and tables on the OpenArchitect switch switching hardware. SYNOPSIS zreg [-p <ppa>] [-w] [-i <index>] [-t <index>] [-k] [-h <hostname>] [-d <level>] [-r 10] <reg> DESCRIPTION zreg allows a user to read and write direct and indirect registers and tables on the resident switch chip.
Page 323
connected (i.e., the one that is on the local PCI bus). -r 10 Sets numeric radix for registers to 10. Default is 16. -d <level> Set the level of debugging output produced by zreg. The default level is 1. Setting the debug level higher produces more output.
zrld NAME zrld – ZNYX redirector daemon SYNOPSIS zrld [-d <level>] [-p <port>] [-f] DESCRIPTION zrld is used for remote management of OA/HA applications. OA/HA applications capable of remote management include zlc, ztats, zlmd. zrld only allows requests from hosts listed in /etc/rcZ.d/zrld_trusted_hosts. OPTIONS -d <level>...
zsnoopd NAME zsnoopd - IGMP Snooping daemon for the OpenArchitect switch. SYNOPSIS zsnoopd [-d <level>] [-f] [-h <hostname>] [-p <ppa>] [-r <sec>] [-t <sec>] [-u <sec>] [-v <vlan_id>] DESCRIPTION zsnoopd is run after the network interfaces are created and initialized with zconfig, and started with ifconfig(1M).
Page 326
Run zsnoopd in foreground. Default is to run it in background. -h <hostname> Connect to remote host <hostname>. -p <ppa> Start zsnoopd on switch <ppa>. Default is 0. -r <sec> Time to wait, in seconds, before removing a port with no router multicast traffic.
Page 327
zpeer peer state query The output from the above command during the boot process would be “unhealthy” The -a option can be used to display a complete listing of all state and priority information and internal information that can be used for debugging. Here is example output from the -a option. Local/Write Peer/Read priority...
zspconfig NAME zspconfig - configure and start surviving partner SYNOPSIS zspconfig [-d <level>] [-p <directory_path>] [-u <dhcp_interface>] [-c <dhclient.conf>] [-t <timeout>] [-s] [-v] -f <file> DESCRIPTION zspconfig is used to configure and start the Surviving Partner software. With the -f option a configuration file is provided that completely describes the network setup and desired behavior of all of the switches participating in the Surviving Partner.
Page 329
finding a Surviving Partner to retrieve configuration information from. Only valid with the –u option. Do not ask for confirmation. Run from a script. Prints the current version of zspconfig. -f <file> The provided <file> is used as input to configure the Surviving Partner.
Page 330
A sibling_addresses statement is required for each VLAN created with the zconfig commands. The two addresses in the list indicate there are two switches in the Surviving Partner group. The first address 10.0.0.30 and 11.0.0.30 are assigned to the switch on which the configuration is being run.
Page 331
monitor_only - Ports that are monitored but do not have a virtual address managed on them. They will not have their links brought down temporarily during a failover scenario. These ports are only monitored. If a problem occurs on this type of link it will cause a failover scenario.
Page 332
interconnect, it is not a bonding driver enabled port, and therefore should be setup as an interconnect port type. To accomplish this, the zre ports are listed to avoid conflicting port types. Note that a single line cannot contain both zhp and zre definitions. Therefore RAINlink: zhp1, zre1..4 does not work and the definition zre1..8 is equivalent.
Page 333
Coordination between the data and fabric switches can be enabled by setting the board_synchronization_mode. Possible modes are “off” and “basic”. Board synchronization is off by default. When board synchronization is put into basic mode, HA events on the base switch are coordinated with the HA events on the fabric switch. The behavior of board synchronization is dependent on the failover_mode.
Page 334
files run the Surviving Partner. This is a list of all configuration and script files: /etc/rcZ.d/S70Surviving_partner The main startup script that starts the Surviving Partner by running zconfig, ifconfig, zlmd and vrrpd. zspconfig prompts the user to run this script. This file can be saved with zsync to automatically start the Surviving Partner at switch boot.
zstack NAME zstack - Configures the OpenArchitect switch stacking. SYNOPSIS zstack [-h <host_name>] [-d <level>] [-a] [-t] [{-f <file>} | <configuration>] DESCRIPTION zstack combines multiple switch fabric chips into a single virtual switch. zstack must be run before any other switch configuration. Specifically it must be run before zconfig. zstack is typically run from an S20stack script prior to the S50xxx scripts.
Page 337
semicolon-delimited statements. Each statement specifies an action to take on a stack. A stack is a group of ports on a single switch fabric chip. Actions include stack creation, stack port association, stack configuration and stack control. Comments, spaces and new lines are ignored. Comments begin with the # character and include characters through the next new line.
Page 338
semicolon are two zre_lists separated by an equal sign. The first is the list of virtual port names, the second is the physical port names. The assignment is done in order, and there must be an equal number of ports in each list. Wild cards may be used in the zre_lists. See below. Stack port association syntax for a base switch: stack0: zre0..11 = zre0..11;...
Page 339
Stack Control Statements Finally after creating the stack, associating the ports, and setting the stack configuration, the stack can be enabled using one of the Stack Control statements. The following stack control statements are supported. enable; The enable statement turns on stacking that has been previously configured. This statement cannot be made until configuration is complete.
ztats NAME ztats − Display statistics and information about switch SYNOPSIS ztats [-d <level>] [-i <unit>] | [-m <port>] | [-v <vlan id>] | [-t <tgid>] | [-v] DESCRIPTION ztats displays MIB counters for a selected physical port, trunk group or VLAN. It can also display information about the configuration of the switch and bridge to the PCI bus or the Vital Product Data memory.
zsync NAME zsync − Saves changes to the flash. SYNOPSIS zsync [-c][-f][<dir_or_file>] zsync [-c][-f][-t <file>] zsync [-c][-f][-z] zsync [-c][-l] DESCRIPTION zsync is used to save a snapshot of the current file system to flash ROM. By default, zsync creates a compressed tar image of the files that have changed and saves the image in the flash ROM.
Page 342
OPTIONS Save files to the custom overlay -t <file> Read files to be saved from a tar file. Zero the overlay area. Do not confirm with user and do not warn if saving failed. Exit code can be examined to determine success or failure. <dir_or_file>...
ztmd NAME ztmd – traffic management daemon which accepts messages from traffic filtering and quality of service applications and sets up hardware. SYNOPSIS ztmd [-d <level>] [-p <port>] [-f] [-i <pid>] [-o <pid>] [-a <addr>] [-l] DESCRIPTION ztmd listens for messages on a multicast port. These messages describe packet filters and queuing disciplines that are to be installed in the switch hardware.
Page 344
SEE ALSO zqosd, iptables(8), tc(8), zfilterd Ethernet Switch Blade User's Guide release 3.2.2j page 344 Downloaded from www.Manualslib.com manuals search engine...
brctl(8) NAME brctl - Bridge and Spanning Tree Protocol administration. SYNOPSIS brctl [options] DESCRIPTION brctl is used to set up, maintain, and display the bridge configuration in the Linux kernel. brctl is a standard command included with Linux bridge support including Rapid Spanning Tree Protocol (RSTP).
Page 346
makes the interface a port of the bridge. This means that all frames received on the interface will be processed as if destined for the bridge. Also, when sending frames on the bridge, the interface will be considered as a potential output interface. For the OpenArchitect switch, <interface>...
Page 347
setmaxage <bridge> <time> sets the bridge's maximum message age to <time> seconds. setpathcost <bridge> <port> <cost> sets the port cost of the port to <cost>. This is a dimensionless metric. The path cost is set to 100 for all OpenArchitect switch ports by default.
Appendix C Intelligent Platform Management Interface The Ethernet Switch Blade provides Intelligent Platform Management Interface (IPMI) support. IPMI circuitry provides: The communication channel between the Baseboard Management Controller (BMC) and the • CPU for management. Data storage, SDRR, FRU, access. •...
M States Deactivation request (e.g. hot swap ejector opened) Deactivation granted by ShMC Unexpected loss of communication between FRU and ShMC Table C.1.: IPMI M States Peripheral Management Controller Functional Support The following IPMI commands are implemented in version 1.00: PMC Controller Support Command Code...
Sensor Reading Example This is an example of how to structure a command and response to determine a sensor value. In this example, a GetSensorReading command is sent from BMC (address 20h in this example), to the switch in slot 2 (geographical address is B2h) to read the temperature sensor, which is assigned to sensor number 60h.
Standard IPMI Response: GetSensorReading Byte Data Field Description rqAddr netFn/Lun check1 rsAddr seq no command completion code sensor reading 1Bh -> 27 Celsius degree optional data C0h scanning is enabled byte optional data byte optional data byte checksum2 Table C.4: GetSensorResonse Only scanning is supported and enabled for the optional bytes.
Structure of Standard IPMI Commands: From BMC to PMC Structure of Standard IPMI Commands BMC - PMC Byte Data Field Description rsAddr <slot’s IPMB addr> netFn/Lun <netFn> check1 <chksm1> rqAddr <sw_id> seq no <seq> command <cmd> optional data byte <arg1> optional data <argN>...
Event Generator The PMC’s event generator is disabled until it receives a SetEventReceiver command from BMC for Event Receiver’s slave address and LUN. If the event generator is enabled, PMC reports significant events to the BMC asynchronously. The standard IPMI platform event message format is used. IPMB Event message format Structure of event messages sent to the BMC by PM device (PMC) is shown below.
Version 2.x supports three FRU Inventory Device Commands: Get FRU Inventory Area Info Read FRU Data Write FRU Data The spare SEEPROM space is allocated as follows: Spare SEEPROM Space Allocation Space for Start address End address Size Notes 0x3FF 0x400 (1Kbytes) 0x400 0x13FF...