Yuval Lifshitz
Yuval Lifshitz
> @yuvalif do we have any upgrade test coverage to validate these changes? if not, maybe this backport is not worth the risk/effort according to Ali, the dev effort is...
> I missed the message of Casey and so I already implemented a solution in [3f2702b](https://github.com/ceph/ceph/commit/3f2702bffa2c58daf21a2a4e05c85bf3929c554b) if we decide that we want to go on with the solution, I can...
> > @cbodley can we make the change with only manual testing? > > @yuvalif manual testing for 18.2.5 won't help catch regressions in later point releases. we generally only...
require special migration code and testing
the address sanitizer gived errors regarding memory leaks in the internal object in the amqp library. it is probaly a cleanup bug where we do not call the proper cleanup...
@Svelar please note that was major refactoring done for the kafka client code (rgw_kafka.cc). And, we should probably perform similar refactoring on the amqp code. would be great if we...
> Hi @yuvalif, glad to hear that amqp part would also be refactored. Actually I am not familiar with these modules' code, the reason why I opened this PR is...
> > regarding the race condition, the usage of these iterator is done from the managers thread. the only multithreaded access is between the calls in the main loop of...
@Svelar this is an important fix. are you still working on this PR?
@Svelar did you run the new code under ASAN? did it find any issues?