ECS: Upgrade Planning

• Operating System (OS) and Fabric upgrade
• OS Quickfit [may also be seen as OS Hotfix] and Fabric upgrade
• Fabric only upgrades
• Patch upgrade

Upgrade Tools

1. Service Console: Primary tool used to run health checks and the upgrade
2. xDoctor: The primary tool used to check and maintain the ECS appliances health
3. Codes [OS/Fabric]: The packages used to perform updates to the ECS

Upgrade Concerns:

1. The upgrade total time depends on how many nodes are in the cluster (every node takes approximately 45 minutes to be upgraded) and all nodes are rebooted during the upgrade.
2. Review the upgrade preparation guide and prepare any questions we may have as a result and ensure that your environment is protected against all the known concerns mentioned in the document.
3. If the upgrade will be done over SCG, ensure to get at least two nodes, including the master node, available directly on SCG.
4. If there are multi-step upgrades, all zones must be upgraded to the intermediate release first and then left at that intermediate release for ~24 hours before beginning the next upgrade jump.
5. Expect I/O timeouts on all nodes when one node is being upgraded, when a node that is being upgraded enters maintenance mode, all the other nodes in the cluster experience an I/O time-out for approximately 30 to 90 seconds.
6. Upgrades of VDCs that are in replication groups together are not recommended to be done in parallel. The risk of a failure on a single VDC is low and having a replication group can help to mitigate that risk, upgrading multiple VDCs in a replication group in parallel may impact this risk mitigation.

Upgrade Steps

Operating System and Fabric

1. Install the latest releases of Service Console and xDoctor [10 minutes]
2. Generate the cluster.ini file and update the DataNodes file (where needed) [15 minutes]
3. Run a pre-upgrade health check [varies according to the size of the VDC and current code, approximately 30 minutes]
4. Move the upgrade files into the /tmp/Upgrade directory [negligible]
5. Initiate the upgrade from the master node to upgrade node 2
   1. Runs upgrade setup [20 minutes]
   2. Installs Service Console on node 2 to assist with upgrade of all the remaining nodes in the VDC [negligible]
   3. Upgrades and reboots node 2 [45 minutes]
6. Connect to node 2 and continue the upgrade
   1. Upgrade and reboots the remaining nodes [45 minutes for each node]
   2. Runs upgrade tear down [20 minutes]
7. Run upgrade post health checks [varies according to the size of the VDC and current code, approximately 30 minutes]

OS Quickfit/Hotfix + Fabric Upgrade

1. Install the latest releases of Service Console and xDoctor [10 minutes]
2. Generate the cluster.ini file and update the DataNodes file (where needed) [15 minutes]
3. Run a pre-upgrade health check [varies according to the size of the VDC and current code, approximately 30 minutes]
4. Move the upgrade files into the /tmp/Upgrade directory [negligible]
5. Perform the OS Quickfit [30 minutes]
6. Perform the Fabric upgrade
   1. Runs upgrade setup [20 minutes]
   2. Upgrades each node [20 minutes for each node]
   3. Runs upgrade tear down [20 minutes]
7. Run upgrade post health checks [varies according to the size of the VDC and current code, approximately 30 minutes]

Fabric Only Upgrade

1. Install the latest releases of Service Console and xDoctor [10 minutes]
2. Generate the cluster.ini file and update the DataNodes file (where needed) [15 minutes]
3. Run a pre-upgrade health check [varies according to the size of the VDC and current code, approximately 30 minutes]
4. Move the upgrade files into the /tmp/Upgrade directory [negligible]
5. Perform the upgrade
   1. Runs upgrade setup [20 minutes]
   2. Upgrades each node [20 minutes for each node]
   3. Runs upgrade tear down [20 minutes]
6. Run upgrade post health checks [varies according to the size of the VDC and current code, approximately 30 minutes]

Patch [Steps are the same as Fabric Only upgrades]