This document supports the newest release and the previous release of FC Switch OEM firmware qualified for use with the DYNIX/ptx V4.4.x operating systems.
ATTENTION The OEM vendor, Brocade Communications Systems, Inc., has implemented a scheme to differentiate its OEM software distribution channels, such as IBM, from its direct-market distribution channels, such as their own website for software downloads. The version name of this particular OEM release has a different form than previous releases from this vendor. The first character in this new form is a lowercase "a" instead of the usual uppercase "V." There is no content difference in the distribution types, but the OEM type can be installed on an OEM-sourced FC Switch without needing to install a new license key. A direct-sourced distribution would require a specific license or else all ports on the switch will be disabled following an upgrade and reboot.
When an IBM-supported FC Switch is queried with the version command during a telnet session, the switch's operating system should respond with a version label that shows that the operational firmware came from the OEM distribution, such as "a2.2.1a."
DYNIX/ptx system tools such as ptx/INSTALL and other listing utilities do not accommodate a lowercase character in the first position of a label. System tools will substitute an uppercase V preceding the version numbers of this software package, for instance, "V2.2.1." When inspecting a log file such as /etc/versionlog for installation history, the version numbers are the significant information.
See Table 1-1 for the latest recipe for the best set of SAN FC firmware to run on a given operating system.
The family of FC Switches used on IBM xSeries 430 and IBM NUMA-Q servers includes the following models:
The 16-port switch IBM 2109 S16, also nomenclated as SilkWorm 2800
The 8-port switch IBM 2109 S08, also nomenclated as SilkWorm 2400
These switches are supplied by an OEM, Brocade Communications Systems, Incorporated.
The switch models are identical except for their nomenclature and exterior color scheme. Switches with IBM nomenclature are all black and switches with the Brocade SilkWorm nomenclature are have purple accents on the front panel.
New system and add-on orders are shipped with IBM 2109 switches. Older systems being upgraded to a new software release are most likely to be using the SilkWorm-nomenclated models.
ATTENTION On DYNIX/ptx V4.4.x, all FC switches must have a license key enabled. Otherwise, EES events, which are captured by in-band communication with the FC switch, will not be logged.
Additionally, zoning cannot be enabled without a valid license key. While the zoning commands are enabled by the presence of a ZONING license key, IBM does not support general usage of the zoning commands on xSeries 430 or IBM NUMA-Q servers by customers. Because these commands can easily disable a system if used incorrectly, IBM requires deployment of the zoning feature only by IBM Professional Services.
The name of the switch:
Must have the first character be 'a-z' or 'A-Z' or a numeral.
Can only contain the characters 'a-z', 'A-Z', '0-9', or the underscore (_). No other punctuation is allowed.
Cannot exceed 19 characters in length.
Additionally, it is suggested that switch names be unique in a cascaded environment.
See Set an FC Switch Name for procedures to set or change the name of a switch.
Version a2.2.1a is the June 2001 release of firmware for the Fabric Operating System of the IBM 2109 and SilkWorm 2000-family of FC Switches for DYNIX/ptx V4.4.10, V4.4.9, V4.4.8, and V4.4.7.
The xSeries 430 and IBM NUMA-Q servers use most, but not all, of the new features present in the new release. See the discussion in "Changes Since the Previous Release," Section 5.2.2
This New Release section contains the following subsections:
To install the software distribution files, see "Install V2.x FOS Software for FC Switches," Section 5.3.
To download firmware to the switches, see "Download to IBM 2109 and SilkWorm 2000 Firmware," Section 5.4.
When running on an IBM 2109 or SilkWorm 2000 FC Switches, version a2.2.1a of this OEM firmware is compatible with the following:
DYNIX/ptx V4.4.10, V4.4.9, V4.4.8, and V4.4.7 and other members of system-software release sets as listed in Table 1-1, "'Best Recipe' FC-Firmware Release Sets for DYNIX/ptx V4.4.x System," in Chapter 1.
SilkWorm 1000 FC Switches running, V1.6c3 firmware only, in the following circumstances:
All hardware versions of IBM xSeries 430 and IBM NUMA-Q servers
All hardware versions of IBM 2109 and SilkWorm 2000 FC Switches
The non-OFC versions of the FC Bridge
The non-OFC versions of the LP6000 FC Host Adapter board
The non-OFC versions of the LP6000 FC Host Adapter board
The IBM FAStT200TM Storage Server
The HDSTM 5800TM Storage Subsystem
EMC® Symmetrix® Storage Subsystems with only SA Director cards and running at least V5264.34.32 firmware.
EMC Symmetrix Storage Subsystems with only FC Director cards and running at least V5265.11.11 firmware.
EMC Symmetrix Storage Subsystems with mixed SA Director and FC Director cards and running at least V5265.11.11 firmware.
CLARiiON® DASS Storage Subsystems running at least LIC firmware V9.55.01.
When this release is installed, a new directory, /ibm_2109 is added in the /usr/ssw/fw/fc_switch directory to contain the new binary file. The new binary file that contains the release software is named fc_sw_ibm2109.bin.
The existing directory named /brocade2000 and its binary file fc_sw_217.bin remain.
The xSeries 430 and IBM NUMA-Q servers do not use all of the features present in this release. The following new commands are used by those servers:
The following telnet commands have been added to the admin account. All of these commands are supported with help command man pages.
(See the Fiber Channel Switch Administration Guide for task-oriented procedures that use these commands.)
"HEALTHY/OK." Every status contributor is under the thresholds set in the Status Policy file.
"Marginal/Warning." One or more of the contributors has exceeded the threshold for this status. The contributing factors will be listed along with the status report.
"DOWN/FAILED." One or more of the contributors has exceeded the threshold for this status. The contributing factors will be listed along with the status report.
The seven contributors that make up the set, and their default values for each status, are:
Contributor |
Marginal Status |
Down Status |
Internal Switch State |
Offline, or Online diagnostic testing |
Faulty |
Faulty Ports |
1 |
2 |
Missing GBICs |
0 |
0 |
Power Supplies |
1 |
2 |
Temperatures |
1 |
2 |
Fans |
1 |
2 |
Port Status |
0 |
0 |
In the following example, parenthese enclose the possible range of values and the current value is shown inside a pair of brackets. To change any current value, make an entry at the end of the line and press RETURN. For example:
sw14:admin>switchStatusPolicySet
To change the overall switch status policy parameters . . . Note that the value, 0, for a parameter, means that it is NOT used in the calculation. ** In addition, if the range of settable values in the prompt is (0..0), ** the policy parameter is NOT applicable to the switch. ** Simply hit the Return key. The minimum number of FaultyPorts contributing to DOWN status: (0..16) [2] FaultyPorts contributing to MARGINAL status: (0..16) [1] MissingGBICs contributing to DOWN status: (0..16) [0] 2 MissingGBICs contributing to MARGINAL status: (0..16) [0] 1 Bad PowerSupplies contributing to DOWN status: (0..2) [2] Bad PowerSupplies contributing to MARGINAL status: (0..2) [1] Bad Temperatures contributing to DOWN status: (0..5) [2] Bad Temperatures contributing to MARGINAL status: (0..5) [1] Bad Fans contributing to DOWN status: (0..4) [2] Bad Fans contributing to MARGINAL status: (0..4) [1] Down PortStatus contributing to DOWN status: (0..16) [0] Down PortStatus contributing to MARGINAL status: (0..16) [0] Policy parameter set has been changed Committing configuration......done sw14:admin>
Default values are set when firmware is downloaded and the switch is rebooted. They are also reset when the configDefault command is issued.
sw14:admin>switchStatusPolicyShow
The current overall switch status policy parameters:
Down Marginal
----------------------------------
FaultyPorts 2 1
MissingGBICs 0 0
PowerSupplies 2 1
Temperatures 2 1
Fans 2 1
PortStatus 0 0
The help man page for this command explains how the calculation is performed. It also explains the ramifications of setting thresholds to 0 (zero).
The state of this mode is reported in the switchShow command.
The following changes are tracked and reported to the Error Log. Inspect the Error Log with the errShow command.
Support for the following telnet commands has been enhanced. Both of these commands are documented in the help command man pages.
No new commands have been added to the Control Panel interface of the 16-port switches, IBM 2109 S16 and SilkWorm 2800.
Most problems described in this section have workarounds that allow the system to recover and continue operating without loss of data. The problems without workarounds do not jeopardize data I/O integrity and are mostly internal administrative functions within the switch.
The following problems have been found in a2.2.1a. See "FC I/O Subsystem-Level Problem Reports," Chapter 2 for possible I/O subsystem-level problems involving FC Switches.
If the Ethernet IP address field is set to "none," the switch's name will be displayed as an empty string following the next reboot.
Workaround: Set the Ethernet IP address and reboot the switch.
Sometimes, in a cascade, changes in the enabled zoning configurations are not propagated to other switches after a switch reboot.
Workaround: Assuming that the system is at the single-user level and there is no I/O activity, do the following sequence of telnet commands on each switch in the cascade that did not get updated with the changed configuration:
Take each switch offline (switchDisable) except the switch where the zone configuration change was first created and enabled.
Clear any existing zone information from the database of each switch (cfgClear).
Save that state in each switch (cfgSave).
Put each switch back online(switchEnable).
Reboot each switch (reboot.). As each rebooted switch comes up, the new enabled zoning configuration will be automatically propogated to it from the source switch (now the Principal switch) over the ISLs of the cascade.
Periodically, an "fcpCommand: iu 0x10ec4f60 iu_status 0x18 (timed out)" error will be reported.
Workaround: Ignore this error report.
Sometimes, when a RAID device logs in, the inquiry data (symbolic name) is missing from the name server information table and does not show in the nsShow command.
Workaround: Remove the device from the switch and then add it back in.
Sometimes the firmwareDownload command takes a very long time to complete.
Workaround: Power cycle the switch and perform another download operation.
When the online diagnostics of the DYNIX/ptx V4.4.x operating system requests status information using the SES protocol, sometimes more pages than the expected Page 0 are returned.
Workaround: No known workaround.
Even though extensive instructions on the following switch management features appear in the hard copy User Guides that accompany the 2109 FC Switches, xSeries 430 or IBM NUMA-Q servers do not support nor use commands associated with:
Quick Loop functionality
Fabric Watch switch management
SNMP switch management
WEB TOOLS switch management
Of the commands available from the telnet interface of the IBM 2109 and SilkWorm 2000 FC Switches, xSeries 430 or IBM NUMA-Q servers do not use nor support the following specific commands:
agtcfgDefault
aliasShow
tempShow
gbicShow
msConfigure
Of the menu items that appear on the Control-Panel interface of the 16-port switches, IBM 2109 S16 and SilkWorm 2800, xSeries 430 and IBM NUMA-Q servers do not use nor support the following commands:
Disable Device Probing
Disable RLS Probing
Suppress Class F Traffic
Frame Collection
Always Send RSCN
Disable FAN Frames
VC Priorities
VC Multicast
VC Class 3
VC Class 2
VC Link Ctl
Temperature
The Fabric Operating System software for IBM 2109 and SilkWorm 2000 FC Switches can be installed at anytime:
For DYNIX/ptx V4.4.10 systems, use the distribution CD-ROM labeled "DYNIX/ptx V4.4.10 OS and Layered Products Software, Vol. 1, December 2001, Rev. A."
For DYNIX/ptx V4.4.9 systems, use the distribution CD-ROM labeled "DYNIX/ptx V4.4.9 OS and Layered Products Software, Vol. 1, August 2001, Rev. B."
For DYNIX/ptx V4.4.8 systems, use the distribution CD-ROM labeled "DYNIX/ptx V4.4.8 OS and Layered Products Software, Vol. 1, August 2001, Rev. C."
For DYNIX/ptx V4.4.4, V4.4.6, and V4.4.7 systems with Static Hardware Partitioning deployed, use the distribution CD-ROM labeled "Fibre Channel Software for DYNIX/ptx V4.4.4, V4.4.6, and V4.4.7."
Insert the distribution CD-ROM into the CD-ROM drive in the Bootbay connected to Quad 0 of the host system.
Log in to the host as root.
ATTENTION There are two families of FC Switch firmware listed during the installation process, one for SilkWorm 1000s and one for IBM 2109 or SilkWorm 2000s. Be sure to read the descriptions and load only the correct switch software package from the installation menus.
Use the ptx/ADMIN® menu system and the ptx/INSTALL utility to install the appropriate software files as described next:
For DYNIX/ptx V4.4.10, V4.4.9, V4.4.8, and V4.4.7, install the "V2.2.1" FC Switch software into the /usr/ssw/fw/fc_switch/ibm_2109 directory.
For DYNIX/ptx V4.4.4, V4.4.6, and V4.4.7 systems that are deploying Static Hardware Partitioning, install the IBM 2109 and SilkWorm 2000 FC Switch software named "V2.1.3" into the /usr/ssw/fw/fc_switch/brocade2000 directory.
ATTENTION In the ptx/INSTALL tool and all other DYNIX/ptx listing utilities, the software package named "V2.2.1" contains the OEM software that will be reported by the FC Switch version command as "a2.2.1.a" after downloading it.
Verify that all directories in the paths and the binary files have read and write permissions for all. If any one does not, the related firmware download command will fail.
Remove the distribution CD-ROM.
To download and flash the software to firmware, use the procedure in the next section.
ATTENTION The following procedures will require the system to be at the single-user level during the firmware upgrade.
The following information is required before beginning the downloading procedure:
Switch symbolic name, such as "sw9" (switchName)
Severity level for displaying error messages (errDisplayFilter)
Collect this list before beginning the upgrade.
Software downloading is by TCP connection between the Switch and the host system containing the source file. Before attempting to download software to a Switch, use the following checklist to make sure that the host system was properly configured when the system or Switch was initially installed. If these host-system files are not configured as described, you may get the "Permission denied" and the "rcmd fails" error messages. The configuration procedures are described in Section 8.5 of the Fibre Channel Subsystem Installation. Here is the checklist:
From the host system, verify that it can communicate with each Switch to be upgraded. Issue the following command:
username$ /etc/ping Switch_IP_address
Do upgrade procedures with the host system at the single-user level.
The following procedures are performed from the PTX Console window via a remote login (telnet) over the Ethernet LAN connected to the IBM 2109 or SilkWorm 2000 FC Switch to be upgraded.
Warn all users and then use the shutdown command to bring the host system down to the single-user level .
Enable basic networking and the inetd daemon from the single-user level with the following commands:
# /etc/ifconfigall 2
# /etc/inetd
Verify network operation. For example:
# netstat -in
Name Mtu Network Address Ipkts Opkts Odrop
loop 0 <Link> 0 0 0
loop 0 127/8 127.0.0.1 0 0 0
pe0 1502 <Link>00.00.bc.0f.04.ee 57 12 0
pe0 1502 138.95.103.24 138.95.103.77 57 12 0
If the Switch is not on the same side of a network router as the host, use the route command on the host system. For example:
# route fill
add net default: gateway 138.95.103.215
Initiate a telnet login session with the first Switch to be upgraded and log in as admin. For example:
(Host_prompt): telnet switch_IP_address
(Switch_name) login: admin
Password: admin_password
sw3:admin>
If this switch is one of a cascaded pair making up a fabric, determine its current domain ID at this time. The domain ID is the first item in the field named "Switch ID" in the following display:
sw3:admin> fabricShow
Switch ID Worldwide Name Enet IP Addr FC IP Addr Name
------------------------------------------------------------------
1: fffc41 10:00:00:60:69:00:0a:12 192.168.1.2 0.0.0 >"sw2"
2: fffc42 10:00:00:60:69:00:01:b4 192.168.1.3 0.0.0 "sw3"
After the firmware has been upgraded, the ID must be restored to this value before returning the switch to service.
ATTENTION The ">" in the display indicates the Principal switch in the fabric.
Put the Switch into its offline state with the following command:
sw3:admin> switchDisable
Observe that the yellow LEDs of all the ports flash steadily, indicating all ports unavailable.
Download the new software into the Switch's flash memory with the following firmwareDownload command:
ATTENTION The syntax of double quotes and commas to separate each argument is important.
The username in the following command is the account set up on the host system for administering Switches.
sw3:admin> firmwareDownload "host_IP_address", "username",
"/usr/ssw/fw/fc_switch/ibm_2109/fc_sw_ibm2109.bin"
1159644+195092+876080
writing flash 0 ................
writing flash 1 ................
download complete
value 0 = 0x0
sw3:admin>
Make the new firmware effective at this time by rebooting the Switch.
sw3:admin> reboot
Observe that the green LEDs of all the cabled ports glow steadily, indicating ports available.
ATTENTION During the reboot, the telnet session will time out and be closed by the remote host.
Start a new telnet session with the Switch and log in as admin.
Verify the currently-executing firmware version. For example:
sw3:admin> version
VxWorks version: 5.3.1
Firmware version: a2.2.1a
Made on: Fri Aug 25 15:47:31 PDT 2000
Flash: Fri Aug 25 15:47:58 PDT 2000
Boot PROM: Thu Jun 17 15:20:39 PDT 1999
where:
Since the reboot cycle enables the Switch, you must disable it again before issuing a configuration command:
sw3:admin> switchDisable
value = 0 = 0x0
Issue the following command to ensure that all configuration values are reset to the factory defaults.
sw3:admin> configDefault
value = 0 = 0x0
Turn on in-order frame delivery:
sw3:admin> iodSet
Committing configuration ...done.
Turn off the dynamic load-sharing for routing:
sw3:admin> dlsReset
Committing configuration ...done.
The configure command used in this step presents each configurable parameter one line at a time. The range of values will be shown in the parentheses following the colon of the prompt. Following the range of values, the current value will be shown in the square brackets. Enter a new value at the end of the line and press the Return key. To leave the value as is, make no entry and press the Return key.
Configure only the "Fabric Parameters." At the end of the list of Fabric Parameters, press Ctrl-D to write the changes and exit from the command. IBM NUMA-Q servers do not use the Virtual Channel and Arbitrated Loop Parameters or System services listed in the rest of the command's parameters. They will retain their default settings.
Set only these two Fabric Parameters to a value of 1:
VC Encoded Address Mode
Disable Translative Mode
ATTENTION If the VC Encoded Address Mode value is left as "0," then the associated parameter Disable Translative Mode will not be displayed by the configure command. The VC Encoded Address Mode must be set to "1" to activate the Disable Translative Mode parameter.
Leave all other Fabric Parameters unchanged.
For example:
sw3:admin> configure
Configure...
Fabric parameters (yes, y, no, n): [no] yes
Domain: (1..239) [1]
BB credit: (1..16) [16]
R_A_TOV: (4000..12000) [10000]
E_D_TOV: (1000..5000) [2000]
Data field size: (256..212) [2112]
Non-SCSI Tachyon Mode: (0..1) [0]
Disable Device Probing: (0..1) [0]
VC Encoded Address Mode: (0..1) [0] 1
Disable Translative Mode: (0..1) [0] 1
Per-frame Routing Priority: (0..1) [0]
Virtual Channel (yes, y, no, n): [no] Crtl-D
Committing configuration ...done.
sw3:admin#
Issue the configure command again and verify that the values are changed. Exit the command with a Crtl-C command after the Fabric Parameters have been reviewed.
For instance:
ATTENTION Enabling the VC Encoded Address Mode and Disable Translative Modes causes the range for domain IDs to be restricted and changed to 0-31.
sw3:admin> configure
Configure...
Fabric parameters (yes, y, no, n): [no] yes
Domain: (0..31) [1]
BB credit: (1..16) [16]
R_A_TOV: (4000..12000) [10000]
E_D_TOV: (1000..5000) [2000]
Data field size: (256..212) [2112]
Non-SCSI Tachyon Mode: (0..1) [0]
Disable Device Probing: (0..1) [0]
VC Encoded Address Mode: (0..1) [1]
Disable Translative Mode: (0..1) [1]
Per-frame Routing Priority: (0..1) [0]
Virtual Channel (yes, y, no, n): [no] Crtl-C
No changes.
sw3:admin>
Set the domain ID:
ATTENTION For serviceability reasons, we recommend that all switches have unique domain IDs even though the switches may be in opposite fabrics. This technique allows for easier correlation from the system dumpconf listings to the physical switches.
The default domain ID for the V2.x Fabric Operating System family is 1 after a firmware download or a specific "Reset to Default" action such as the configDefault command.
If this one switch makes up the entire fabric, set it to an ID different than the switch in the opposite fabric, within the range 0-31.
If this switch is cascaded to another IBM 2109 or SilkWorm 2000-family switch, match the value determined back in Step 5. This method keeps the new system dumpconf listings the same as the old listings.
If this switch is cascaded to a SilkWorm 1000-family switch in a mixed fabric, then make its new domain ID unique and within the range 0-31.
Issue the configure command again.
If the default value is the desired value, then use Crtl-C command to quit and exit the command with no changes. For example:
sw3:admin> configure
Configure...
Fabric parameters (yes, y, no, n): [no] yes
Domain: (0..31) [0] Crtl-C
No changes.
If the default value inside of the square brackets is not the desired value, then enter the preferred value and use Crtl-D to write the new value into flash and exit the command. For example:
sw3:admin> configure
Configure...
Fabric parameters (yes, y, no, n): [no] yes
Domain: (0..31) [1] 3
BB credit: (1..16) [16] Crtl-D
Committing configuration ... done.
Restore other switch configuration values such as site-specific passwords, usernames, and site-specific identities at this time:
Use the passwd command to recreate any non-default passwords for the username accounts.
Use the passwd command also to rename any of the username accounts.
Use the errDispSet command to restore the error-severity level for reporting to the display panel of a 16-port switch.
Use the agtcfgSet command to restore the variables for sysDescr:, sysLocation:, and sysContact: only.
Power cycle the Switch at this point to enable it and to put the restored values into effect:
ATTENTION Since Switches do not get power cycled very often in a customer production environment, it is a good idea to take advantage of the upgrade opportunity to make a switch go through its entire POST sequence.
Repeat this process for each IBM 2109 or SilkWorm 2000 switch in the system.
When all Switches in a system have been upgraded, boot the system back up to the multiuser level with the following command:
# init 6
Verify that all cabled port LEDs on the upgraded Switches show a steady green, indicating that each upgraded Switch has re-established FC links to the system's host adapters and to FC-ported storage subsystems that are connected to each Switch.