Disruptive software upgrade
This task provides step-by-step
instructions about how to perform a disruptive software upgrade using the
CLI.
The IBM Total SAN Volume Controller only supports concurrent code
upgrades. To ensure that a code
upgrade is coordinated across all nodes in the cluster, it is necessary for
the nodes to be able to communicate with each other across the fibre-channel
SAN. However, some users may prefer to perform a disruptive code upgrade.
The following procedure documents how to quiesce I/O to the SAN before performing
a concurrent code upgrade to ensure that there is no I/O in progress during
the upgrade.
Steps
Perform the following steps to complete the disruptive software
upgrade process:
- Stop any host applications and unmount the filesystems that are using
storage that is being managed by the SAN Volume Controller. If your hosts are being shutdown
then this will occur as the host is shutdown, otherwise it will be necessary
to do this manually on each host. This step will ensure that hosts will stop
issuing I/O operations and that any data in the filesystem caches is flushed.
- Shutdown the cluster by issuing the svctask stopcluster command. This command will stop the SAN Volume Controllers from issuing I/O to back-end
controllers and will flush data from the SAN Volume Controller cache.
- Re-zone the switch so that the SAN Volume Controller nodes are in one zone. Ensure that
this zone does not include a host HBA or a back-end controller (keep the old
switch configuration so it can be restored at step 6).
This step isolates the SAN Volume Controller from the rest of the SAN.
-
Power on all the SAN Volume Controller nodes and wait for them to reform
a cluster.
 | Note: Because the IBM Total Storage SAN Volume Controller has been isolated
from the back-end storage you will get some error logs indicating that this
has occurred. |
- Perform the software upgrade in the same manner as for a concurrent code
upgrade.
-
Restore the original switch configuration.
- Clear any error logs produced at step 4 indicating
that back-end storage is unavailable. Check that all back-end storage is now
online and accessible to the SAN Volume Controllers.
- Remount filesystems and start host applications.
Related topics