> On April 7, 2017, 10:40 a.m., Peter Bacsko wrote:
> > Are we sure LOG.trace() is good? I would vote for DEBUG level.

Done.


> On April 7, 2017, 10:40 a.m., Peter Bacsko wrote:
> > core/src/main/java/org/apache/oozie/service/ZKLocksService.java
> > Line 59 (original), 57 (patched)
> > <https://reviews.apache.org/r/58172/diff/1/?file=1684293#file1684293line60>
> >
> >     Are these used by the tests? If not, modify to private.

Yes, it's used in multiple other class files in the same package.


> On April 7, 2017, 10:40 a.m., Peter Bacsko wrote:
> > core/src/main/java/org/apache/oozie/service/ZKLocksService.java
> > Lines 164 (patched)
> > <https://reviews.apache.org/r/58172/diff/1/?file=1684293#file1684293line167>
> >
> >     One thing that that makes me wonder - what if the lock already exists? 
> > What happens to newLockEntry? I think ZK client takes care of it and then 
> > gets reaped by a background thread. But still, at this point, we know that 
> > we don't need the newly created lock - is there any way to get rid ot it?
> 
> Peter Bacsko wrote:
>     Quickly opened the JavaDoc of Curator. Looks like we cannot explicitly 
> destroy the lock. I think it's worth adding a single comment that having 
> newLockEntry is not a problem, like:
>     
>     
>     // We can't destoy newLockEntry and we don't have to
>     
>     This way we let other developers know that this is not a result of us 
> being forgetful:)

Done.


> On April 7, 2017, 10:40 a.m., Peter Bacsko wrote:
> > core/src/test/java/org/apache/oozie/service/TestZKLocksService.java
> > Lines 559 (patched)
> > <https://reviews.apache.org/r/58172/diff/1/?file=1684294#file1684294line559>
> >
> >     It's good that you created tests, but do we test stuff like logging? Is 
> > it worth it? I rarely see tests that verify logger calls. Let's discuss 
> > this in private.

As this patch is explicitly created for improving logging, I think it's worth 
to test exactly the logging behavior.


- András


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/58172/#review171325
-----------------------------------------------------------


On April 4, 2017, 10:40 a.m., András Piros wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/58172/
> -----------------------------------------------------------
> 
> (Updated April 4, 2017, 10:40 a.m.)
> 
> 
> Review request for oozie and Peter Bacsko.
> 
> 
> Repository: oozie-git
> 
> 
> Description
> -------
> 
> Sometimes when Oozie tries to acquire lock [via 
> ZooKeeper](https://github.com/apache/oozie/blob/master/core/src/main/java/org/apache/oozie/service/ZKLocksService.java#L144)
>  or 
> [in-memory](https://github.com/apache/oozie/blob/master/core/src/main/java/org/apache/oozie/service/MemoryLocksService.java)
>  we don't get enough information whether the action succeeded or not. Also 
> other details like retry count, timeout applied, etc. are not logged. 
> Enhancing logging here would benefit debugging of lock handling within 
> `XCommand` instances.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/oozie/service/MemoryLocksService.java 
> 2ab2abc34831daf75bf050b1e4c907afcb627978 
>   core/src/main/java/org/apache/oozie/service/ZKLocksService.java 
> 83790cf27772b106d39779d2ec5ccfa27ce52ef4 
>   core/src/test/java/org/apache/oozie/service/TestZKLocksService.java 
> d04f04e80a5a7ae9051d3a94feece82a9d460f67 
> 
> 
> Diff: https://reviews.apache.org/r/58172/diff/1/
> 
> 
> Testing
> -------
> 
> New unit test created, existing ones that matter remained running.
> 
> 
> Thanks,
> 
> András Piros
> 
>

Reply via email to