[SERVER-36495] Cache pressure issues during recovery oplog application - MongoDB Jira

XML

Word

Printable

JSON

During recovery oplog application we don't advance the oldest timestamp. This can pin a lot of data in the cache. Resulting symptoms include

slow recovery due to resulting cache churn
growth of lookaside (cache overflow) table and corresponding file WiredTigerLAS.wt
in extreme cases cache-full hangs were observed. This was believed fixed with ~~SERVER-33191~~, but we should be alert for new occurrences of this issue.

is duplicated by

SERVER-38698 Forgot to setOldestTimestamp when crash recovery in MongoDB 4.0

is related to

SERVER-38089 Fatal error 40292 due to missing oplog entry at the start time stamp

related to

SERVER-35191 Stuck with cache full during rollback

SERVER-33191 Cache-full hangs on 3.6

SERVER-34938 Secondary slowdown or hang due to content pinned in cache by single oplog batch

SERVER-34941 Add testing to cover cases where timestamps cause cache pressure

SERVER-35339 Complete recovery failure after unclean shutdown

SERVER-36238 replica set startup fails in wt_cache_full.js, initial_sync_wt_cache_full.js, recovery_wt_cache_full.js when journaling is disabled

(3 related to)