/
Configuration Manager for Jira Upgrade issues

Configuration Manager for Jira Upgrade issues

Audience: Jira Data Center administrators upgrading Appfire’s Configuration Manager for Jira (CMJ) or other apps via UPM.

Symptoms

  • Jira becomes slow or unresponsive during an app upgrade in Universal Plugin Manager (UPM).

  • All or many apps appear to unload/reload; users may see “Preparing Jira” or experience failed page loads.

  • Upgrade appears to stall until a Jira restart is performed; after restart, the upgraded app shows as installed and running.

Root cause (what’s happening)

During app upgrades, Jira’s plugin framework unloads and re-initializes apps. Under load or on busy nodes, this can temporarily starve resources and trigger a wider plugin restart cycle. This is a Jira-side behavior; CMJ and most apps cannot override it. In some environments, a clean restart is required for the upgrade to settle.

Goal: Minimize impact and avoid cluster-wide disruption while upgrading.

Before you begin

  • Plan a short maintenance window and notify users. Prefer low-traffic periods.

  • Confirm Jira application health: all nodes green, indexes healthy, no elevated error rates in logs.

  • Have access to the hosting platform to restart services if needed.

 

Highlevel upgrade procedure (Data Center)

  1. Announce a brief maintenance window and ensure low load.

  2. Enable UPM Safe Mode to minimize plugin re-initialization overhead.

  3. Perform the app upgrade. If the UI hangs or reports “Pending restart,” restart Jira to complete the upgrade cleanly.

Recommended upgrade procedure (Data Center)

  1. Quiesce and drain one node:

    • Remove the node from the load balancer or put it in maintenance so end-user traffic avoids it.

    • Verify no long-running background tasks (indexing, reindex, bulk operations) on this node.

  2. Enter UPM Safe Mode on the drained node:

    • From Manage apps, choose “Enter safe mode” to disable non-system apps, reducing the plugin system churn during upgrade.

  3. Upgrade CMJ on that node:

    • Install/upgrade the target app version in UPM.

    • If the UI becomes unresponsive or the upgrade reports “Pending restart,” perform a controlled Jira restart on this node.

  4. Exit Safe Mode and validate:

    • Confirm the app is Enabled and “Running.”

    • Check application logs for plugin errors or repeated restarts.

  5. Return the node to the load balancer and repeat on remaining nodes one at a time.

Verification checklist

  • Manage apps shows CMJ and its bundled modules enabled: Configuration Manager Core Bundle, Configuration Manager for Jira, Integrity Check for Jira, Power Admin, CMJ SPI Bundle.

  • No recurring plugin reload loops or ERROR entries related to plugin initialization in logs.

  • Node(s) rejoined cluster successfully; user traffic normal.

Best practices to reduce downtime

  • Prefer maintenance windows with minimal concurrent user activity.

  • Upgrade one node at a time in Data Center; avoid cluster-wide simultaneous upgrades.

  • Use UPM Safe Mode to limit non-essential plugin activity during upgrades.

  • Ensure adequate JVM heap and system resources to prevent resource contention during plugin reloads.

FAQ

Because the plugin framework may not fully re-initialize all components under load. A restart gives the system a clean state to finalize the upgrade.

Zero-downtime cannot be guaranteed. Following the Data Center rolling approach (drain one node, Safe Mode, upgrade, validate, reintroduce) minimizes user impact.

Typically no. Standard UPM upgrades are supported. If you encounter repeated initialization issues, a clean reinstall (removing CMJ and its bundled modules, then installing the desired version) may help.

Related guidance

Atlassian UPM Safe Mode and Plugin Troubleshooting: https://confluence.atlassian.com/upm/entering-safe-mode-781192636.html