-
Task
-
Resolution: Done
-
Major
-
None
-
None
-
None
We are sometimes seeing traces which show "write flush complete" logs within the timeout period for close. These are likely associated with a subsequent write event, after the write has taken place, but do not extend the grace period for close.
This happens under load and likely means that the timeout has already triggered, we should tolerate this event and extend the grace period as the code intends.
- is cloned by
-
MAISTRA-1860 Check delayed close expiry to prevent possible early close
-
- Closed
-
bravery300 There is no reliable reproducer for this issue, we just have the logs which show that the close was triggered earlier than the 1s as required.