belugabehr
belugabehr
I've been looking at this question quite a bit in-depth. The biggest issue in my mind is trying to cleanup after oneself in failure scenarios. I think the answer is:...
@ctubbsii Yes. Tablet appears to be a 'hot' lock. I am actually looking at that class but it's a bit of a beast. Anyway we can remove these external `Tablet`...
@keith-turner Thanks for the input. So, the downside with your proposed solution is that the locking becomes very granular. That is to say, if a method needs to be called...
Also, typically, I use `isHeldByCurrentThread()` in `private` scoped methods to ensure the lock is not lost with a future change.
You can check out my work on the matter here: https://github.com/belugabehr/accumulo/tree/TabletLock
For this particular PR it's still not clear to me why `CommitSession.java` needs to lock (block) the entire tablet to keep this internal count.
Hello, This moves all the code into a new package, so I can set my logger to more verbose just for this one package to get better understand where the...
I too am struggling on this. Nothing I do seems to get spot-bugs to load the XML file. I cannot figure out why without additional logging. Finally figured it out....
Hello, Thank you for the feedback. I agree a unit test is appropriate here. Ideally, the hashcode implementation should return the same value for the same contents regardless on the...
I have a unit test that confirms the behavior, but it does require AVRO-4061 to be merged first.