-
Type: Bug
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: None
-
Fully Compatible
-
ALL
-
v4.4
-
Sharding EMEA 2022-06-13
-
(copied to CRM)
SERVER-60142 introduced a uassert in case the deadline on the migration destination manager is hit while waiting for a range to be cleared up. This specific change prevents the balancer from getting blocked and must be backported to v5.0 and v4.4 (the original backports didn't include this part of the fix).
When the balancer gets blocked waiting for a range deletion, a thread on the recipient shard shows the following stack trace:
10.000 │ ├▽ bool mongo::Interruptible::waitForConditionOrInterruptUntil<...> 5.000 │ │├▽ mongo::CollectionShardingRuntime::waitForClean 5.000 │ ││ ▽ mongo::MigrationDestinationManager::_migrateDriver 5.000 │ ││ ▽ mongo::MigrationDestinationManager::_migrateThread 5.000 │ ││ ▽ ??? 5.000 │ ││ ▽ ??? 5.000 │ ││ ▽ ??? 5.000 │ ││ ▽ clone
- backports
-
SERVER-60142 Shard can migrate on top of orphans after filtering metadata was cleared
- Closed