Upgrade process flow
The upgrade procedure by using the installation toolkit is done in several
phases. If you are doing manual upgrade, you can use this process flow as a
reference.
Upgrade phases
- Upgrade precheck
- Phase 1: Upgrade of all non-protocol nodes
- Phase 2: Upgrade of protocol nodes (except SMB and CES components)
- Phase 3: Object version sync, SMB upgrade, and CES upgrade on protocol nodes; License package upgrade on all nodes; and File audit logging message queue enabling at cluster level.
- Upgrade post-check
Upgrade precheck
Important: It is highly recommended to run the upgrade precheck while you are planning
to upgrade, before the actual upgrade procedure. Doing this allows you adequate time to address any
issues flagged during the upgrade precheck.
As part of the upgrade precheck, the following tasks are performed by the installation
toolkit.
- Determine the installed IBM Spectrum Scale packages across the cluster on all nodes.
- Compare the versions of the installed packages with the versions in the repository of the
packages you want to upgrade to.
In a mixed operating system cluster, the comparison is done with the package repository applicable for the operating system running on the respective nodes.
- Determine which packages need to be upgraded on each node.
- For every package that needs to be upgraded or installed, dependency check is done to ensure that all dependencies are met. The upgrade procedure throws an error and exits, if any dependency is not met.
Phase 1
| 1 | Check the cluster health. |
| Perform following steps on each non-protocol node, one node at a time. | |
| 2 | Check the node health. |
| 3 | Unmount file systems and shut down GPFS. Attention: At this point, I/O from this node stops.
|
| 4 | Upgrade core GPFS. |
| 5 | Build GPFS portability layer (GPL). |
| 6 | Upgrade file audit logging. Install if file audit logging is enabled in the cluster configuration but not installed. |
| 7 | Upgrade performance monitoring. |
| 8 | Upgrade GUI. |
| 9 | Start up GPFS and mount the file systems. Attention: At this point, I/O from this node resumes.
|
| 10 | Check the node health. |
| These steps are repeated on the remaining non-protocol nodes, one node at a time. After all of the non-protocol nodes are upgraded, phase 1 is complete. |
Phase 2
| Perform following steps on each protocol node, one node at a time. | |
| 1 | Check the node health. |
| 2 | Suspend CES on the node. Attention: At this point, protocol I/O from this node
stops.
|
| 3 | Unmount file systems and shut down GPFS. Attention: At this point, I/O from this node stops.
|
| 4 | Upgrade core GPFS. |
| 5 | Build GPFS portability layer (GPL). |
| 6 | Upgrade file audit logging. Install if file audit logging is enabled in the cluster configuration but not installed. |
| 7 | Upgrade object. Note: Object services are stopped on the node and all object packages are
upgraded on the node. Object services are started on all nodes in phase 3 of the upgrade
procedure.
|
| 8 | Upgrade NFS. |
| 9 | Upgrade performance monitoring. |
| 10 | Upgrade GUI. |
| 11 | Start up GPFS and mount the file systems. Attention: At this point, I/O from this node resumes.
|
| 12 | Resume CES on the node. Attention: At this point, protocol I/O from this node
resumes.
|
| 13 | Check the node health. |
| These steps are repeated on the remaining protocol nodes, one node at a time. After all of the protocol nodes are upgraded, phase 2 is complete. |
Phase 3
| 1 | Perform object version sync from one of the protocol nodes and start object services on all
protocol nodes. Attention: At this point, object I/O resumes in the whole
cluster.
|
| 2 | Suspend CES on the first half of nodes. |
| 3 | Stop SMB on the 1st half of nodes. Attention: At this point, SMB I/O on this
half of nodes stops and CES IPs on these nodes are migrated to the nodes in the other
half.
|
| 4 | Upgrade SMB on the 1st half of nodes. |
| 5 | Suspend CES on the 2nd half of nodes. Attention: At this point, SMB I/O stops
on the whole cluster for a brief period.
|
| 6 | Stop SMB on the 2nd half of nodes. |
| 7 | Start SMB on the 1st half of nodes. |
| 8 | Resume CES on the 1st half of nodes. Attention: At this point, SMB I/O resumes
on this half of nodes.
|
| 9 | Upgrade SMB on the 2nd half of nodes. |
| 10 | Start SMB on the 2nd half of nodes. |
| 11 | Resume CES on the 2nd half of nodes. Attention: At this point, SMB I/O resumes
in the whole cluster.
|
| 12 | Upgrade CES on all protocol nodes, one node at a time. |
| 13 | Upgrade license package on all nodes, one node at a time. |
| 14 | Enable file audit logging message queue on the cluster. |
| 15 | Check the cluster health. |
| After these steps are done, the installation toolkit automatically runs upgrade post-check. |
Upgrade post-check
As part of the upgrade post-check, the following tasks are performed by the installation
toolkit.
- Check the health of the nodes.
- Compare the versions of the installed packages with the versions in the package repository.
In a mixed operating system cluster, the comparison is done with the package repository applicable for the operating system running on the respective nodes.
- If any package is not upgraded, the installation toolkit prompts you to rerun the upgrade procedure.
You must complete the upgrade to the new code level to take advantage of the new functionality. For more information, see Completing the upgrade to a new level of IBM Spectrum Scale.
Related concepts: