-
Bug
-
Resolution: Unresolved
-
Low
-
None
-
6.15.4
-
5
-
Severity 2 - Major
-
2
-
Issue Summary
When running Confluence in a clustered database and performing a failover, Synchrony will fail to reestablish connection to the database upon database engine rejoining.
All Synchrony transaction will fail after fail over is completed.
Environment
Verified on Aurora and Postgres RDS running external Synchrony.
Steps to Reproduce
1. Run Confluence with external Synchrony on a RDS Postgres
2. Open up collaborative editor with two sessions
3. Fail over RDS Postgres
4. Attempt to make changes in collaborative editor and observe Synchrony connection cannot be established
Expected Results
Synchrony automatically establishes connection after fail over.
Actual Results
Synchrony does not automatically establish connection after fail over.
Workaround
Restarting the Synchrony node will force the connection to be reestablished.
- mentioned in
-
Page Failed to load
| Form Name | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
[CONFSERVER-58260] Synchrony fails to reestablish connection after a fail over when configured with a clustered database
After the failover was finished, we were able to see 3 connections still using the reader Aurora node and this exception was thrown in the browser console when trying to edit collaboratively.
11:06:50.741 VM219:1251 warn reinit Caught BatchUpdateException for insert into "EVENTS" ("history", "rev", "partition", "sequence", "event") values (?, ?, ?, ?, ?) Error: Caught BatchUpdateException for insert into "EVENTS" ("history", "rev", "partition", "sequence", "event") values (?, ?, ?, ?, ?) at new Xj (eval at <anonymous> (http://confl-loadb-3z05bq85m17r-1494529760.us-west-2.elb.amazonaws.com/s/2a828c72c2d24752e89e028afd5fe809-CDN/en_US/7901/5add2ffb254089f9b2b4da47cac4a1fe5d074b7a/1388f45a017d5e2ce90810891801350e/_/download/contextbatch/js/_super/batch.js?locale=en-US:409:128), <anonymous>:489:26) at Yj (eval at <anonymous> (http://confl-loadb-3z05bq85m17r-1494529760.us-west-2.elb.amazonaws.com/s/2a828c72c2d24752e89e028afd5fe809-CDN/en_US/7901/5add2ffb254089f9b2b4da47cac4a1fe5d074b7a/1388f45a017d5e2ce90810891801350e/_/download/contextbatch/js/_super/batch.js?locale=en-US:409:128), <anonymous>:490:75) at eval (eval at <anonymous> (http://confl-loadb-3z05bq85m17r-1494529760.us-west-2.elb.amazonaws.com/s/2a828c72c2d24752e89e028afd5fe809- ....
Connection tuning seems to mitigate this problem. Setting reasonable values for the following properties (or env vars) will help synchrony reconnect to a database after a failover event
We need to test collaborative editing by setting configurations to different values and failing over and document the recommended settings