curator-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Cameron McKenzie <mckenzie....@gmail.com>
Subject Re: CURATOR-161
Date Wed, 20 May 2015 22:04:01 GMT
I've had a look at the master branch and see that this code has been fixed
by using a guaranteed delete on the mutex. So, that explains why it was
failing in 3.0 and not previously. My git knowledge isn't super strong, so
I have no idea why master may not be merging into 3.0 I'm afraid. I had a
quick look through the log and it doesn't appear that 3.0 has been merged
into master which would the the only thing I could think of as to why
master would think it didn't need to be merged into 3.0.

Maybe one of the other devs has an idea?
cheers

On Thu, May 21, 2015 at 3:55 AM, Jordan Zimmerman <
jordan@jordanzimmerman.com> wrote:

> I just pushed CURATOR-3.0-temp. I started fresh with master and merged
> CURATOR-215, CURATOR-160 and CURATOR-186. This may be the only way to fix
> things. I’m not sure what happened.
>
> -Jordan
>
>
>
> On May 20, 2015 at 3:29:38 AM, Cameron McKenzie (mckenzie.cam@gmail.com)
> wrote:
>
>   Thanks Jordan,
> I've done the license stuff and that seems to have worked OK, but on
> rerunning all the tests after merging in 3.0 I have found an issue with one
> of the tests.
>
> In TestLeaderSelector the testLeaderNodeDeleteOnInterrupt test is failing.
> I'm not sure why it's suddenly failing now when it hasn't been previously.
>
> It seems to be that the test code blocks the main leadership thread, then
> main test code interrupts this blocked leadership thread and expects
> leadership to be interrupted. The problem is that in the LeaderSelector
> doWork() method, if the thread it is executing in is interrupted then in
> its finally block when it goes to release the mutex, the release never
> occurs because the thread is in an interrupted state. When it attempts to
> do the actual delete on the ZK client it throw an InterruptedException
> immediately which is silently ignored by the LeaderLatch.
>
> Is this a known problem that hasn't been merged into the 3.0 branch?
> cheers
>
> On Wed, May 20, 2015 at 3:38 PM, Jordan Zimmerman <
> jordan@jordanzimmerman.com> wrote:
>
>>  However, you can just run:
>>
>>  mvn license:format
>>
>>  Yourself and it should be fine. maybe try that
>>
>>
>>
>> On May 19, 2015 at 9:22:32 PM, Cameron McKenzie (mckenzie.cam@gmail.com)
>> wrote:
>>
>>  Jordan,
>> I noticed that in CURATOR-217 you fixed up the license headers of a bunch
>> of the files that were committed as part of CURATOR-161. I don't want to
>> merge CURATOR-161 back to the 3.0 branch without the headers as I think it
>> will break the build. So, what's the best way forward? Do you just want to
>> merge CURATOR-217 back to the 3.0 branch given that it already contains all
>> of the CURATOR-161 changes?
>> cheers
>>
>> On Wed, May 20, 2015 at 9:23 AM, Jordan Zimmerman <
>> jordan@jordanzimmerman.com> wrote:
>>
>>>  https://github.com/apache/curator/pull/82
>>>
>>>
>>>
>>> On May 19, 2015 at 4:18:36 PM, Cameron McKenzie (mckenzie.cam@gmail.com)
>>> wrote:
>>>
>>>  Thanks Jordan,
>>> Will have a look at the PR.
>>> cheers
>>>
>>> On Wed, May 20, 2015 at 9:17 AM, Jordan Zimmerman <
>>> jordan@jordanzimmerman.com> wrote:
>>>
>>>>  I’ll have a look. Also, I already started on integrating watcher
>>>> removal as I have strong ideas on it. Here’s the current PR:
>>>> https://github.com/apache/curator/pull/80
>>>>
>>>> <https://github.com/apache/curator/pull/80>
>>>>  - <https://github.com/apache/curator/pull/80>JZ
>>>>
>>>>
>>>> On May 19, 2015 at 4:06:03 PM, Cameron McKenzie (cammckenzie@apache.org)
>>>> wrote:
>>>>
>>>>  Guys,
>>>> If anyone's got a spare few minutes could they please review the final
>>>> changes to CURATOR-161. If we're happy with the implementation then I'll
>>>> merge into the 3.0 branch and then start work on getting the recipes to
>>>> use
>>>> the watcher removal stuff.
>>>> cheers
>>>> Cam
>>>>
>>>>
>>>
>>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message