[GitHub] flink issue #4764: [FLINK-7757] [checkpointing] Introduce resource guard for...

2017-10-18 Thread StefanRRichter
Github user StefanRRichter commented on the issue: https://github.com/apache/flink/pull/4764 Thanks for the reviews @aljoscha @bowenli86 and @GJL . I addressed all points and will merge this now. ---

[GitHub] flink issue #4764: [FLINK-7757] [checkpointing] Introduce resource guard for...

2017-10-03 Thread bowenli86
Github user bowenli86 commented on the issue: https://github.com/apache/flink/pull/4764 I was commenting on the scenario described in `What is the purpose of the change`. In that case, we want to either fail s2 or remove the lock for s2 as you did. Since we are required to support par

[GitHub] flink issue #4764: [FLINK-7757] [checkpointing] Introduce resource guard for...

2017-10-03 Thread StefanRRichter
Github user StefanRRichter commented on the issue: https://github.com/apache/flink/pull/4764 @bowenli86 Can you go into detail what kind of problem you have in mind? This class does not work like a semaphore, i.e. `acquire` never truly blocks. It is just incrementing a count and does

[GitHub] flink issue #4764: [FLINK-7757] [checkpointing] Introduce resource guard for...

2017-10-03 Thread bowenli86
Github user bowenli86 commented on the issue: https://github.com/apache/flink/pull/4764 The implementation of having s2 `acquire()` a lock while s1 still holds that lock seems problematic to me. Why not making it a `tryAcquire()`, as that in java semaphore? ---

[GitHub] flink issue #4764: [FLINK-7757] [checkpointing] Introduce resource guard for...

2017-10-03 Thread StefanRRichter
Github user StefanRRichter commented on the issue: https://github.com/apache/flink/pull/4764 CC @aljoscha ---