ConnectALL Sync Fails to Update Rally Artifacts Due to Jira Search Indexing Latency

Products

• ConnectALL
• ConnectAll On-Prem
• ConnectAll SaaS

Issue/Introduction

Users may experience a scenario where updates made to a Sprint value in Jira do not synchronize to the corresponding artifact in Rally. Despite the update occurring in the source system, the ConnectALL dashboard may show no sign of failure, and the artifact remains unchanged in the destination.

Environment

ConnectALL Version: 4.0.4, 4.0.5

Source System: Jira

Cause

The root cause is related to how Jira handles search indexing. When an issue field is updated in Jira, the database is updated immediately, but the JQL search index is updated asynchronously. 

If a ConnectALL sync poll occurs during the short delay before the update is indexed, the JQL query used by the adapter (e.g., updated > last_run_time) will not return the modified record. Because the record is not "seen" by the search query at that specific moment, it is skipped for that poll cycle. If retries are disabled or the polling window is very tight, the update may be missed entirely.

Resolution

To resolve this issue, a code change was implemented in ConnectALL 4.1.2 to introduce a configurable "leeway" or buffer time for indexing delays.

• Upgrade ConnectALL: Install or upgrade to ConnectALL version 4.1.2 or later.
• Enable Retries: Ensure that the retry property is not set to zero, as this allows the system to attempt synchronization again if a transient delay occurs.
• Verify Indexing Performance: If the issue persists under heavy load, consult with your Jira administrator to verify the health and speed of the Jira search indexer.
• Logging: For future troubleshooting, detailed indexing logs can be enabled in Jira to correlate sync times with indexing completion.

Additional Information

• ConnectALL logs may show "Found 0 modified records" even if updates were made in Jira, confirming that the items were not yet available to the search API at the time of the poll. 
• Full synchronizations typically bypass this issue as they do not rely on the same incremental "updated after" logic.