Because of the tightly-coupled nature of the nodes in an IRIS FailSafe cluster, you must be careful when performing both routine upgrades—such as installing new operating system releases, patches, or additional software (other than IRIS FailSafe software) or performing hardware upgrades—and IRIS FailSafe upgrades, such as a new release of IRIS FailSafe or a changed configuration file. This chapter describes the upgrade procedures for an IRIS FailSafe cluster.
The major sections of this chapter are as follows:
Three upgrade procedures are presented in this chapter. All upgrades are done using one or more of these procedures, called A, B, and C. Table 7-1 lists the upgrades you may need to perform and the procedure you should use for each of these upgrades. If you need to perform more than one procedure, follow these guidelines:
If you need to perform both procedure A and procedure B, you can use just procedure A: you can perform all of the A and B upgrades during procedure A.
You must complete all of the upgrades that require procedure A or procedure B before beginning procedure C.
If you are replacing a CHALLENGE node with an Origin node while you are upgrading from IRIS FailSafe 1.1 to IRIS FailSafe 1.2, the best approach is to install and configure IRIS FailSafe as if it were the initial installation on a new cluster. The configuration file can be based on the configuration file used in the original cluster. See the section “Overview of Upgrading an IRIS FailSafe Cluster From Release 1.1 to Release 1.2” in Chapter 1 for a list of changes to the configuration file between 1.1 and 1.2.
Upgrade Procedure | Upgrade |
|---|---|
A | Upgrade IRIS FailSafe 1.1 to IRIS FailSafe 1.2 on both nodes |
B | Perform a hardware upgrade on one node |
B (perform once on each node) | Perform hardware upgrades on both nodes |
B | Upgrade IRIX 6.2 or IRIX 6.4 to the latest patch level |
B | Upgrade IRIX 5.3 to IRIX 6.2 on one node (IRIS FailSafe 1.1 is a prerequisite and IRIX 5.3 at the latest patch level or IRIX 6.2 on the other node is a prerequisite) |
B (perform once on each node; the node that doesn't have the latest 5.3 patches must be upgraded first) | Upgrade IRIX 5.3 to IRIX 6.2 on both nodes (IRIS FailSafe 1.1 is a prerequisite and at least one of the nodes must have the latest patches installed) |
B | Install software (other than IRIX and IRIS FailSafe) or IRIX patches that require a reboot on one node |
B (perform once on each node) | Install software (other than IRIX and IRIS FailSafe) or IRIX patches that require a reboot on both nodes |
B | Install software (other than IRIX and IRIS FailSafe) or IRIX patches that don't require a reboot on one node |
B (perform once on each node) | Install software (other than IRIX and IRIS FailSafe) or IRIX patches that don't require a reboot on both nodes |
C | Upgrade from a 1.1 ha.conf file to a 1.2 ha.conf (see “Configuration File Formats and Their Versions” in Chapter 4) |
C | Upgrade from a 1.2 ha.conf file to a 1.2 ha.conf file with changes or additional high-availability services |
Follow the steps below to perform procedure A. The procedure assumes that both nodes are running IRIS FailSafe and are in normal state. Both nodes are rebooted during this procedure.
Enter these commands to shut down IRIS FailSafe on the node you want to upgrade first (host1) and prevent it from restarting automatically:
# chkconfig failsafe off # /etc/init.d/failsafe stop |
The other node, host2, automatically moves into degraded state (providing all high-availability services for the cluster).
On host2, enter this command to verify that it is in degraded state:
# /usr/etc/ha_admin -i ha_admin: Node controller state degraded |
Perform any hardware upgrades on host1.
Use normal software installation procedures (miniroot or inst from IRIX depending upon what you are installing) on host1 to install IRIX 6.2 if desired, IRIS FailSafe 1.1, and other software such as patches and software options.
Reboot host1.
On host2, enter these commands to shut down IRIS FailSafe and prevent it from restarting automatically:
# chkconfig failsafe off # /etc/init.d/failsafe stop |
At this point, no high-availability services are available from either node.
Start IRIS FailSafe on host1 and put it into standby state (providing no high-availability services):
# chkconfig failsafe on # /etc/init.d/failsafe start |
Wait about 90 seconds and confirm that host1 is in standby state:
# /usr/etc/ha_admin -i ha_admin: Node controller state standby |
Move host1 to degraded state:
# /usr/etc/ha_admin -G host1 ha_admin: host1 successfully moved to degraded |
All high-availability services are now being provided by host1. host2 is automatically rebooted.
Perform any hardware upgrades on host2.
Use normal software installation procedures (miniroot or inst from IRIX depending upon what you are installing) on host2 to install IRIX 6.2 if desired, IRIS FailSafe 1.1, and other software such as patches and software options.
Reboot host2.
Enter these commands on host2 to restart IRIS FailSafe:
# chkconfig failsafe on # /etc/init.d/failsafe start |
host1 and host2 automatically move to normal state (providing the high-availability services for which they are the primary node).
Confirm that both nodes are in normal state by entering this command on each node:
# /usr/etc/ha_admin -i ha_admin: Node controller state normal |
Follow the steps below to perform procedure B. The procedure assumes that both nodes are running IRIS FailSafe 1.1 and are in normal state. This procedure upgrades a single node. To upgrade both nodes, perform the procedure twice, once on each node.
Enter this command on the node you are upgrading (host) to put it into standby state (no high-availability services running):
# /usr/etc/ha_admin -s ha_admin: host successfully moved to standby |
The other node moves into degraded state (providing all high-availability services).
Perform any hardware upgrades on host.
Use normal software installation procedures (miniroot or inst from IRIX depending upon what you are installing) on host to install the software you want to install (any software except IRIS FailSafe).
Reboot host if the software you installed requires it. (A date of Jan 1 1970 for the file /unix indicates a reboot is required.)
Check the states of both nodes:
# /usr/etc/ha_admin -i ha_admin: Node controller state normal # /usr/etc/ha_admin -i otherhost ha_admin: Node controller state normal |
otherhost is the hostname of the node you are not upgrading.
If either of the nodes are not in normal state, move them to normal state by entering this command on host:
# /usr/etc/ha_admin -rf ha_admin: host successfully reintegrated |
Both nodes are providing the high-availability services for which they are the primary node.
Follow the steps below to perform procedure C. It assumes that both nodes are running IRIS FailSafe and are in normal state. All of the commands shown can be entered on either node.
Follow steps 1 through 8 in the section “Creating a Configuration File” in Chapter 4 to prepare a 1.1 configuration file and verify it with ha_cfgverify, but not install it.
On the node you want to upgrade first (host1), enter this command to move this node into standby state (no high-availability services running) and the other node (host2) into degraded state (providing all high-availability services):
# /usr/etc/ha_admin -s host1 ha_admin: host1 successfully moved to standby |
host1 is a hostname.
On host1, enter these commands to install the configuration file in /var/ha/ha.conf:
# mv conffile /var/ha/ha.conf # chown root.sys /var/ha/ha.conf # chmod 500 /var/ha/ha.conf # /usr/etc/ha_cfgchksum 0x12a2390e |
Put the node in degraded state (host2) into standby state by entering this command on host2:
# /usr/etc/ha_admin -sf host2 ha_admin: host2 successfully moved to standby |
host2 is a hostname.
At this point, no high-availability services are available.
Immediately put host1 into degraded state by entering this command on host1:
# /usr/etc/ha_admin -rf host1 ha_admin: host1 successfully moved to degraded |
All high-availability services are now being provided by host1.
On host2, enter these commands to install the configuration file in /var/ha/ha.conf.
# mv conffile /var/ha/ha.conf # chown root.sys /var/ha/ha.conf # chmod 500 /var/ha/ha.conf # /usr/etc/ha_cfgchksum 0x12a2390e |
The checksums output by ha_cfgchksum on each node must be identical.
Move both nodes to normal state by entering this command on host2:
# /usr/etc/ha_admin -rf host2 ha_admin: host2 successfully reintegrated |
Both nodes are providing the high-availability services for which they are the primary node.