The Balancer will keep config.migrations documents on shard stepdown / shutdown errors. And return BalancerInterrupted to the mongos, though it isn't stepping down. The mongos will retry the moveChunk command, so it doesn't appear to be a big problem, though it isn't what we intended to happen / really want.
- causes
-
SERVER-46669 moveChunk may succeed but not respect waitForDelete=true if replica set shard primary steps down
- Closed
- is depended on by
-
SERVER-26304 killOp in migration_ignore_interrupts.js is failing to thoroughly kill a moveChunk because now the moveChunk gets retried on ErrorCodes::Interrupted
- Closed
- related to
-
SERVER-25999 Mongos applies errors received from config server as config server errors, rather than a shard the config server calls and returns the error from
- Closed