Below is the sequence with timing and potential impact:
- Remote Proactive Services will perform pre-upgrade health checks for any activity that requires cluster reboots (15-30 minutes); this is usually done just before starting the upgrade. Please note: Pre-Upgrade health checks are also performed during the scheduling phase (up to 2 weeks before the upgrade)
- Remote Proactive Services engineer will make sure all the codes are available (10-20 minutes if the upgrade is done over WebEx or up to 2 hours if Remote Proactive Services will transfer the code over ESRS, also done days prior to the upgrade to ensure actual upgrade starts on scheduled time)
- Before starting the upgrade, Remote Proactive upgrade engineer contacts the point of contact (Customer/CE) to update about pre-upgrade analysis, and if required, get permission to start upgrade.
Rolling upgrade (OneFS, Node Firmware, and Drive firmware): Time in total:
2 hours for clusters with minimum amount of nodes (3-5)
- Remote Proactive upgrade engineer carries out pre-upgrade health check right before the upgrade. 15 minutes
- Upgrade engineer begins the rolling upgrade. For an understanding of what impacts there are during a rolling upgrade, review the Client connections during rolling upgrades section of the OneFS Upgrade Planning and Process Guide available on the EMC Support site. 15 minutes per node
- Upgrade engineer performs post-upgrade health check to ensure the cluster is in good state. 15 minutes
- Please allow extra time for the upgrade to take in case any ad-hoc issues arise during any stage of the upgrade. 30 minutes
Simultaneous OneFS code upgrade: Time in total:
2 hours regardless of node count
- Remote Proactive upgrade engineer carries out pre-upgrade health check right before the upgrade. 15 minutes
- Upgrade engineer kicks off the Simultaneous upgrade. Data access disruption is expected throughout this stage. 30 minutes
- Upgrade engineer performs post-upgrade health check to ensure the cluster is in good state. 15 minutes
- Please allow extra time for the upgrade to take in case any ad-hoc issues arise during any stage of the upgrade. 30 minutes
Rolling reboot or patch install that requires rolling reboot: Time in total:
1.5 hours for clusters with minimum amount of nodes (3-5)
- Remote Proactive upgrade engineer carries out pre-upgrade health check right before the upgrade. 30 minutes
- Upgrade engineer performs reboots of nodes in a rolling manner. For an understanding of what impacts there are during a rolling reboot, review the Client connections during rolling upgrades section of the OneFS Upgrade Planning and Process Guide available on the EMC Support site. 10 minutes per node.
- Upgrade engineer performs post-upgrade health check to ensure the cluster is in good state. 15 minutes
- Please allow extra time for the upgrade to take in case any ad-hoc issues arise during any stage of the upgrade. 30 minutes
Rollback plan:
You should be aware that after you upgrade to a new major version of OneFS, you cannot restore the previous version. OneFS version upgrades consist of re-imaging the kernel and file system and copying user changes from the old file system to the new one. Once the OneFS kernel is re-imaged, there is no mechanism for rolling back to the previous major version
With any upgrade from a cluster that is already running OneFS 8.0. The OneFS code must be committed after the upgrade completes, this should be completed within 2 weeks of the upgrade. During this period there will be the option to roll back to the previous OneFS version and not commit the version that was upgraded to. Be advised, this may only be performed in a simultaneous manner and at the supervision of Isilon Technical Support Services.
Note:
OneFS code upgrade may be performed via rolling or simultaneous upgrade, based on the target version. Node and Drive Firmware upgrades may only be performed only via a rolling upgrade. Refer to the release note for the target OneFS code to verify as to whether a rolling/simultaneous upgrade is supported from current OneFS code.
After the OneFS code upgrade is performed, an SMB roll-up patch (if available) is always recommended to be applied after the upgrade. SMB roll-up patches which will interrupt SMB and authentication services for 15 minutes.
If any OneFS patch that requires a rolling reboot for the cluster is also requested alongside the OneFS code upgrade, the whole upgrade process will need to be extended in order to allow extra reboots to complete.