I am looking at https://jira.mongodb.org/browse/BF-33674. Here jstests/concurrency/fsm_workloads/analyze_shard_key.js test is failing with Hang analyzer killing it after timeout. Issue here is, At recovery time, while applying commitTransaction oplog, it is hitting write conflict continuously until it get resolved. Currently When write conflict happens we increment the number of attempts in logWriteConflictAndBackoff() and keep retrying. When it hits 10k limit we print the log only in case of debug build. In the production build, we don't print any log. Isn't it good idea to print some log even in case of production build with same or higher limit? Otherwise there is no way to find this issue in production build.
- related to
-
SERVER-92178 WiredTiger WriteConflictException should Include the actual namespace or at least the table name that encountered the write conflict
- Backlog