13.5.1 Maintenance Operations on the Non-Global Zone

The non-global zones are monitored and controlled by the Cmdline resource set up to the global zone. Because of this, if operating the RMS or applications on the non-global zone while the global zone RMS is running, this may be detected as an error on the global zone side. Moreover, because of this events such as forced stop of the non-global zone and switchover may occur.

As such, if performing maintenance tasks which include the following operations, follow the maintenance task procedures explained below.

Starting up/stopping the non-global zone
Starting up/stopping the non-global zone RMS
Starting up/stopping the cluster applications being controlled by the non-global zone RMS

As a basic way of thinking, use the maintenance mode on the global zone for maintenance on the non-global zone.

For the following procedures, it is assumed that the global cluster applications that are to undergo maintenance are in Online status. For troubleshooting in situations of some kind of error occurring on the non-global zone and being unable to put the applications into Online status, refer to "13.5.4 Recovery Operation for When an Error Occurs on the Non-Global Zone."

Note

If the virtual IP addresses of fast switching mode or GS/SURE linkage mode are shared in the shared IP zone, cancel the shared state according to the following procedure before stopping the non-global zone.

Shared logical interfaces will be deleted when stopping the non-global zone. Therefore, you need to cancel the state beforehand so that logical virtual interfaces controlled by GLS are not deleted.

Perform this procedure in the global zone.

By using the ifconifg command, check whether the zone setup of the non-global zone name to be stopped (settings of the shared state) is displayed for the logical virtual interfaces (interfaces beginning with "sha") of fast switching mode or GS/SURE linkage mode.
```
# ifconfig -a
sha10:65: flags=1000863<UP,BROADCAST,NOTRAILERS,RUNNING,...
zone zone-a
inet 192.168.100.101 netmask ffffff00 broadcast 192.168.110.255
```

Cancel the shared state if it is set.

# ifconfig sha10:65 -zone
# ifconfig -a
sha10:65: flags=1000863<UP,BROADCAST,NOTRAILERS,RUNNING,...
inet 192.168.100.101 netmask ffffff00 broadcast 192.168.110.255

If GLS logical virtual interfaces are deleted by stopping the non-global zone before performing the procedure by mistake, refer to "Accidentally deleted the virtual interface with ifconfig command" of "Troubleshooting" in the "PRIMECLUSTER Global Link Services Configuration and Administration Guide: Redundant Line Control Function," and then recover GLS logical virtual interfaces as necessary.

The latest release of the following emergency corrections need to be applied to the non-global zone in an OSLC environment.

Solaris 8
- When PRIMECLUSTER 4.1A30 was used in the migration source
  901167-07 or later
  901172-34 or later
  901173-24 or later
  913855-05 or later
  914111-03 or later
  914112-10 or later
  914120-01 or later
  914346-01 or later
  914351-01
  914530-01
  915102-01 or later
- When PRIMECLUSTER 4.1A40 was used in the migration source and using a new PRIMECLUSTER
  901167-07 or later
  901172-34 or later
  901173-24 or later
  913855-05 or later
  914111-03 or later
  914112-10 or later
  914346-01 or later
  914351-02 or later
  914530-01
  915102-01 or later
Solaris 9
- When PRIMECLUSTER 4.1A30 was used in the migration source
  901167-07 or later
  901172-34 or later
  901173-24 or later
  913855-05 or later
  914111-03 or later
  914112-10 or later
  914120-01 or later
  914346-01 or later
  914351-01
  914530-01
  915102-01 or later
- When PRIMECLUSTER 4.1A40 was used in the migration source
  901167-07 or later
  901172-34 or later
  901173-24 or later
  913855-05 or later
  914111-03 or later
  914112-10 or later
  914346-01 or later
  914351-02 or later
  914530-01
  915102-01 or later
- When PRIMECLUSTER 4.2 was used in the migration source and using a new PRIMECLUSTER
  901196-35 or later
  901215-04 or later
  901217-29 or later
  901254-02 or later
  913855-05 or later
  914111-03 or later
  914112-10 or later
  914346-01 or later
  914351-02 or later
  914530-02 or later
  915102-01 or later

If sharing non-global zone images or single-node cluster operations

If sharing non-global zone images and applying a patch to the non-global zones, follow the procedure below. If sharing non-global zone images, patch application by rolling update is not supported.

Migration to maintenance mode
Migrate the corresponding cluster application from the global zone to maintenance mode. It is possible to use GUI and CLI. For the procedure, refer to "7.3 Using maintenance mode" of the "PRIMECLUSTER RMS with Wizard Tools Configuration and Administration Guide."
Performing maintenance tasks on the non-global zone
Perform the necessary maintenance tasks on the non-global zone in the Online state. While the maintenance mode is set on the global zone, it is permitted to use the non-global zone RMS commands (hvcm, hvshut, hvswitch, hvutil, hvdisp, etc.) and to perform the starting up and stopping of the non-global zone as needed.
For example, if applying a patch to the non-global zone, perform the following procedure.
1. Stop the non-global zone. From the global zone, login with zlogin and execute the shutdown command or, as shown below, execute a direct shutdown command from the global zone with a zlogin command.
```
# zlogin zone-a shutdown -y -g0 -i0   *1
```
  *1: "zone-a" is the zone name (It is the same for below).
2. Check that the non-global zone has stopped. To be specific, check that STATUS is "installed" using the zoneadm list command.
```
# zoneadm list -vc
  ID NAME         STATUS  PATH                       BRAND  IP
   0 global           running    /                              native   shared
   - zone-a           installed  /zone-a-system                 native   shared
```
3. Start up the non-global zone under single user mode.
```
# /usr/lib/brand/solaris8/s8_p2v zone-a  (For Solaris 8 Container)
# /usr/lib/brand/solaris9/s9_p2v zone-a  (For Solaris 9 Container)
# zoneadm -z zone-a boot -s
```
4. Apply the patch to the non-global zone.
5. Restart the non-global zone.
```
# zlogin zone-a shutdown -y -g0 -i6
```
Note
- If not sharing non-global zone images, apply the same patch on all of the non-global zones which comprise the cluster system allocated to each global zone.
- Be careful that cluster applications allocated to each global zone are not simultaneously Online.
Non-global zone status recovery
Before canceling maintenance mode, return the non-global zone to the same status it was in at the time it was migrated to maintenance mode. In other words, return it to a status of the non-global zone being started up and the RMS and applications on it being started up. If one manually put the Gds, Gls, Fsystem and other resources into inactive status, also return those resources to active status.
1. Check that the non-global zone has restarted. This status can be checked by acquiring the console with zlogin -C zone-a.
  For single-node cluster operations, the recovery of the cluster applications is performed automatically to the point of startup. Proceed to the next step, "4. Canceling maintenance mode."
```
# zlogin -C zone-a
```
2. After login to the non-global zone, start up the RMS by using the procedure below.
```
# hvcm
```
3. Restore the cluster application on the non-global zone which was in the Online state to the original state.
```
# hvswitch userApp_0   *1
```
  *1: "userApp_0" is the name of the cluster application on the non-global zone.
Canceling maintenance mode
In the global zone, remove the corresponding cluster applications from maintenance mode. It is possible to use GUI and CLI. For the procedures, refer to "7.3 Using maintenance mode " of the "RMS with Wizard Tools Configuration and Administration Guide."

If not sharing non-global zone images

If not sharing non-global zone images, patch application by rolling update is supported.

If applying a patch by rolling update to the non-global zone, perform the following procedure according to the example below.

Example) If applying a patch to the non-global zone on the operational system global zone GZA and standby system global zone GZB using the rolling update.

Execute the following command on GZA and switch the cluster application controlling the non-global zone that is to undergo maintenance over to the standby system GZB.
```
# hvswitch userApp_0 GZBRMS   *1
```
*1: "userApp_0" is the name of the cluster application controlling the non-global zone, whereas "GZBRMS" is the SysNode name.
Migrate the cluster application controlling the non-global zone which is to undergo maintenance to maintenance mode. It is possible to use GUI and CLI. For the procedure, refer to "7.3 Using maintenance mode" of the "RMS with Wizard Tools Configuration and Administration Guide."
Check the status of the non-global zone.
```
# zoneadm list -vc
```
If STATUS is "configured," attach the non-global zone.
```
# zoneadm -z zone-a attach
```
If STATUS is "running," stop the non-global zone. Log in from GZA with zlogin and execute the shutdown command or, as shown below, execute a direct shutdown command from the global zone using the zlogin command.
However, the Cmdline resource that controls the global zones may be in Fault state depending on the settings, continue to perform the steps from now on.
```
# zlogin zone-a shutdown -y -g0 -i0   *2
```
*2: "zone-a" is the zone name (It is the same for below).
Check that the non-global zone has stopped. To be specific, execute the zoneadm list command in GZA and check that STATUS is "installed."
```
# zoneadm list -cv
ID NAME STATUS PATH BRAND IP
0 global running / native shared
- zone-a installed /zone-a-system native excl
```

From GZA, start up under single user mode the non-global zone which is to undergo maintenance.

# /usr/lib/brand/solaris8/s8_p2v zone-a  (For Solaris 8 Container)
# /usr/lib/brand/solaris9/s9_p2v zone-a  (For Solaris 9 Container)
# zoneadm -z zone-a boot -s

Apply the patch to the non-global zone.
If STATUS is "configured" in Step 3, detach the non-global zone.
```
# zoneadm -z zone-a detach
```
If STATUS is "running," restart the non-global zone and start RMS on the non-global zone. If the Cmdline resource which controls the non-global zone becomes Faulted state, wait until the Faulted state of the Cmdline resource is cleared.
```
# zlogin zone-a shutdown -y -g0 -i6   *3
# zlogin zone-a
# hvcm   *4
# hvutil -s userApp_0  *5
```
*3: After this, check that the non-global zone has restarted. This status can be checked by acquiring the console with zlogin -C zone-a.
*4: After login to the non-global zone, start up the RMS and application using the procedure below.
*5: Execute this command when performing the warm standby operation and register resources that can become a standby state to the non-global zone cluster applications. "userApp_0" is the name of a cluster application on the non-global zone.
Cancel the maintenance mode for the corresponding cluster applications. It is possible to use GUI and CLI. For the procedure, refer to "7.3 Using maintenance mode" of the "RMS with Wizard Tools Configuration and Administration Guide."
Apply the patch to the non-global zone allocated to GZB as well. Repeat steps 1) through 8). However, replace "GZA" in the procedure with "GZB" and "GZB" in the procedure with "GZA."

Maintenance to Oracle, NetWorker, and NAS device on the Non-Global Zone

When performing maintenance operations for Oracle, Networker, and NAS device on the non-global zone, it is necessary to temporarily suspend monitoring performed by PRIMECLUSTER or Wizard products.

For maintenance procedures for Wizard products and points of caution, refer to each Wizard product manual.

Applying Corrections to Wizard Products on the Non-Global Zone

If applying emergency corrections to non-global zone Wizard products, put the non-global zone into single user mode and then apply the corrections. At this time, use maintenance mode on the global zone in the same way as for "13.5.1 Maintenance Operations on the Non-Global Zone."

For the procedures and points of caution for applying corrections to Wizard products, refer to each Wizard product manual.