The constructor of Fetcher accepts two timeout values: findNetworkTimeout and getMoreNetworkTimeout. Despite their names, these timeouts govern the entire find/getMore command invocations, not just the network portions. In at least a few cases, these timeouts are derived from a maxTimeMS value, but because they are passed to the Fetcher on their own and not attached to an opCtx, they lose their identities as maxTimeMS values. As a result, if either timeout is hit during the fetching process, a NetworkInterfaceExceededTimeLimit error will be returned (the default defined on RemoteCommandRequest) rather than a MaxTimeMSExpired error, as expected. A potential negative side effect of this would be a process' replica set monitor marking the host as unreachable, when in reality a fetching query just took longer than expected.

An example of this issue can be seen in ShardRemote::_runExhaustiveCursorCommand, which manually extracts the remaining maxTimeMS value from the opCtx and then passes it into a Fetcher.

https://github.com/mongodb/mongo/blob/r7.0.0/src/mongo/s/client/shard_remote.cpp#L303-L310

https://github.com/mongodb/mongo/blob/r7.0.0/src/mongo/client/fetcher.cpp#L200-L201