Do anyone has more information about what happens to “once per cluster” periodic tasks during “zero downtime rolling upgrades”? I’m asking this because, one of my customers reported that scheduled tasks no longer runs during/after “Zero Downtime Rolling Upgrades” until they uninstall/reinstall the app. They are using following procedure to upgrade:
When we upgrade Jira, we do a “rolling update”. We have three nodes of Jira Data Center and when we upgrade using the Zero Downtime Upgrade feature we perform full instance replacement. The sequence of events for our upgrade is:
- Put current cluster of 3 nodes into upgrade mode
- Launch three new nodes one at a time with the new Jira version. This grows the cluster to 6 active nodes.
- Remove the original 3 nodes from the load balancer.
- Five minute cool down to drain any traffic still hitting the old nodes.
- Terminate each of the original 3 nodes.