There have been at least 3 crashes in sharding code caused by use-after-frees or the misuse of references in the 3.1.x cycle: SERVER-19367, SERVER-19839, and SERVER-18815 (the last commit)
We should proactively search for similar issues to avoid costly debugging later.
- is related to
-
SERVER-19367 Segfault establishing shard connection in Chunk::splitMulti
- Closed
-
SERVER-19839 Use-after-free in ShardRegistry::runCommandWithNotMasterRetries
- Closed
-
SERVER-18815 Shard object lifetime shoud be managed by the ShardRegistry
- Closed
- related to
-
SERVER-20559 Race condition in shard registry during concurrent sharding operations
- Closed
-
SERVER-20629 Use-after-free in CatalogManagerReplicaSet::_runReadCommand
- Closed