Traceability matrix links missing in Jira 9 after OSLC sync

ALM Atlassian Jira Software
, , , , , , , ,
1

We’re running Jira 9 with OSLC Connect to sync requirements from external tools. After the latest sync operation, our traceability matrix shows broken links for approximately 180 requirements. The TRS indexer appears to complete successfully, but the links don’t resolve in the matrix view.

Checking the OSLC Connect admin tools, I can see the sync jobs completed without errors. However, when I query the traceability API endpoint, I get inconsistent results across our cluster nodes:

{
  "node1": {"linkedRequirements": 842},
  "node2": {"linkedRequirements": 662},
  "node3": {"linkedRequirements": 842}
}

The bulk operation queue shows 0 pending items, so I’m not sure what’s causing the discrepancy. Has anyone encountered cluster node consistency issues with OSLC traceability links? This is impacting our compliance reporting deadlines.

2

This sounds like a database replication lag issue rather than a cache problem. In Jira Data Center, the traceability links are stored in custom tables that OSLC Connect creates. If your database has read replicas, the bulk sync operation might be writing to the primary while your nodes are reading from replicas that haven’t caught up yet. Check your database replication lag - it should be under 100ms. Also verify that all nodes are pointing to the same database connection pool settings.

3

The connection pool mismatch is definitely part of the issue. Beyond that, OSLC Connect has a background reconciliation job that should detect and fix link inconsistencies. Check if it’s enabled in your configuration - it’s under OSLC Admin > Reconciliation Settings. The job runs every 6 hours by default but you can trigger it manually. Also, for the broken links in the traceability matrix, you might need to rebuild the link index specifically for the affected requirement types.

4

I’ve seen similar sync inconsistencies in clustered Jira 9 environments. The TRS indexer writes to a shared index, but sometimes the cache invalidation doesn’t propagate correctly across nodes. Have you tried forcing a cache clear on all nodes simultaneously? Also check if your OSLC Connect version is compatible with Jira 9 - some older connector versions had cluster awareness issues.