Details
-
Bug
-
Resolution: Unresolved
-
Major - P3
-
None
-
None
-
None
-
None
-
Service Arch
-
ALL
-
Service Arch Prioritized List
Description
wtimeout of zero has a special meaning that it's no timeout. So, logically it has a greater value than anything, but we use numerical comparisons in the places below:
https://github.com/10gen/mongo/blob/231d6367ea3c7670a458ce00252d970979d7f5be/src/mongo/executor/async_rpc_util.h#L74
https://github.com/10gen/mongo/blob/231d6367ea3c7670a458ce00252d970979d7f5be/src/mongo/db/commands.cpp#L509
So, it ends up being treated as a smaller number. As a result, a no timeout setting can be overwritten with a smaller timeout.
Attachments
Issue Links
- is related to
-
SERVER-84443 Bring back no timeout write concern for commands in resharding coordinator
-
- Needs Scheduling
-
-
SERVER-70192 Rewrite uses of sendCommandToShards in sharding ddl coordinator to use async RPC API
-
- Closed
-
- related to
-
SERVER-84356 Make Sharding and User Command Write Concern Timeouts Configurable via Server Parameter
-
- Backlog
-