eolivelli commented on code in PR #3813:
URL: https://github.com/apache/bookkeeper/pull/3813#discussion_r1118406096
##########
stream/distributedlog/core/src/main/java/org/apache/distributedlog/bk/SimpleLedgerAllocator.java:
##########
@@ -382,7 +382,9 @@ private synchronized void completeAllocation(LedgerHandle
lh) {
allocatePath, DistributedLogConstants.EMPTY_BYTES, (int)
version.getLongVersion());
ZKVersionedSetOp commitOp = new ZKVersionedSetOp(zkSetDataOp, this);
tryObtainTxn.addOp(commitOp);
- setPhase(Phase.HANDING_OVER);
+ if (this.phase != Phase.ALLOCATED) {
Review Comment:
It is common to get here when phase is ALLOCATED
see here
https://github.com/apache/bookkeeper/blob/e6722848e0a599f215e68e1340196c9dfe8e906e/stream/distributedlog/core/src/main/java/org/apache/distributedlog/bk/SimpleLedgerAllocator.java#L432
or you can get to here from this path
https://github.com/apache/bookkeeper/blob/e6722848e0a599f215e68e1340196c9dfe8e906e/stream/distributedlog/core/src/main/java/org/apache/distributedlog/bk/SimpleLedgerAllocator.java#L287
maybe it is clearer to not use "!=" but to explicitly enumerate the "phases"
in which you are allowed to transition to `HANDING_OVER` .
Could you please change the condition to a "positive" condition ?
also we could add assertions and fail in case we reach to this point in an
unexpected case
if using the phase is not enough to discriminate the reason why to set
HANDING_OVER maybe we could change the signature of this method and pass the
expected "nextPhase" or something like that
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]