lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-789959546
> I have a WIP solution already. Need to look it a bit more to see if I'm
missing anything and I'll create a PR.
Ok great. I'm looking forward to that.
I'll close this
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-789956789
> To conclude, in my opinion the easiest/lower-risk/lower-overhead change
here would be to switch the forEach() to do the stamped+read-lock combination
for each entry, instead of
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-789608339
> how can we fall into a deadlock ?
Not an easy question. I don't have an answer to that. Perhaps @Vanlightly
could help us?
-
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-789562722
> This map is used in many places throughout the code base and, depending on
the use cases, there can be instances with a lot of entries or many instances
with few entries. Also,
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-788764981
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and u
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-788693090
/pulsarbot run-failure-checks
This is an automated message from the Apache Git Service.
To respond to the message
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-788689691
> The primary goal of having these custom concurrent maps implementation was
actually to avoid memory allocations to the max possible extent :)
>
> Even if these allocation
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-787887121
Here's an example of a suspicious stacktrace from one of the ReplicatorTest
timeouts ([full
threaddump](https://jstack.review/?https://gist.github.com/lhotari/3d2aa04041f86645622
lhotari commented on pull request #9764:
URL: https://github.com/apache/pulsar/pull/9764#issuecomment-787866127
> But I am not sure we can touch so many sensible parts of the codebase in
one single patch.
> We could break things without knowing and it will be hard to track down to
the c
