devinbost commented on pull request #10758:
URL: https://github.com/apache/pulsar/pull/10758#issuecomment-864319361
@eolivelli I moved the `pendingAddEntries.add(addOperation)` calls to after
we're done making changes on the `addOperation` in the different code branches.
What happens if this block in `[ManagedLedgerImpl].createComplete(..)` is
called before the thread processing `pendingAddEntries` has had a chance to
process all the operations (for example, if there was backpressure)?
``` if (pendingAddEntries.isEmpty()) {
STATE_UPDATER.set(this, State.ClosedLedger);
} else {
STATE_UPDATER.set(this, State.WriteFailed);
}
```
(https://github.com/apache/pulsar/blob/380cf924f81a906759811b7a5e861ca66bb2c7ab/managed-ledger/src/main/java/org/apache/bookkeeper/mledger/impl/ManagedLedgerImpl.java#L1403)
Seems like a race... The state could be set to `State.WriteFailed` when in
reality the `addOperation` just hasn't been processed yet. If they're set to
WriteFailed, what happens? Do they get automatically reprocessed? I wonder if
there could be an edge case there.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
