You are viewing a plain text version of this content. The canonical link for it is here.
Posted to builds@apache.org by sebb <se...@gmail.com> on 2011/09/12 18:19:50 UTC
Continuum - Commons Pool unit test is stuck (again)
Commons Pool has been running for over 20 hours:
http://vmbuild.apache.org/continuum/buildResults.action?projectId=98&projectGroupId=16
Please could someone kill the build?
Thanks.
BTW, is there a way to limit the time Continuum allows tests to run?
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by sebb <se...@gmail.com>.
On 13 September 2011 12:27, Brett Porter <br...@apache.org> wrote:
>
> On 13/09/2011, at 6:41 PM, sebb wrote:
>
>> On 13 September 2011 00:21, Brett Porter <br...@apache.org> wrote:
>>>
>>> On 13/09/2011, at 3:27 AM, sebb wrote:
>>>
>>>>
>>>>>>
>>>>>> BTW, is there a way to limit the time Continuum allows tests to run?
>>>>>
>>>>> Done… but it will probably leave any dead builds behind.
>>>>
>>>> Is that a system-wide setting?
>>>
>>> It's a setting for the builds on that schedule. Since commons was on a separate schedule it hadn't been set, but the main one was.
>>
>> What's the setting now please?
>>
>> Continuum does not seem to allow even read-only access to such settings.
>
> It's an overly generous 7200 (2 hours)… more for the default builds as the Directory ones go on quite long when triggered.
That's fine - certainly much better than unlimited as it was before.
> - Brett
>
> --
> Brett Porter
> brett@apache.org
> http://brettporter.wordpress.com/
> http://au.linkedin.com/in/brettporter
>
>
>
>
>
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by Brett Porter <br...@apache.org>.
On 13/09/2011, at 6:41 PM, sebb wrote:
> On 13 September 2011 00:21, Brett Porter <br...@apache.org> wrote:
>>
>> On 13/09/2011, at 3:27 AM, sebb wrote:
>>
>>>
>>>>>
>>>>> BTW, is there a way to limit the time Continuum allows tests to run?
>>>>
>>>> Done… but it will probably leave any dead builds behind.
>>>
>>> Is that a system-wide setting?
>>
>> It's a setting for the builds on that schedule. Since commons was on a separate schedule it hadn't been set, but the main one was.
>
> What's the setting now please?
>
> Continuum does not seem to allow even read-only access to such settings.
It's an overly generous 7200 (2 hours)… more for the default builds as the Directory ones go on quite long when triggered.
- Brett
--
Brett Porter
brett@apache.org
http://brettporter.wordpress.com/
http://au.linkedin.com/in/brettporter
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by Stefan Bodewig <bo...@apache.org>.
On 2011-09-13, sebb wrote:
> On 13 September 2011 10:18, Stefan Bodewig <bo...@apache.org> wrote:
>> On 2011-09-13, sebb wrote:
>>> In case it helps anyone else, I updated the pom as follows:
>>> <artifactId>maven-surefire-plugin</artifactId>
>>> <configuration>
>>> <!-- Don't allow test to run for more than 30 minutes -->
>>> <forkedProcessTimeoutInSeconds>1800</forkedProcessTimeoutInSeconds>
>>> This kills the test; however the output only reports the test class,
>>> not the individual method, and the surefire report file is empty.
>>> To get better info, add @Test(timeout=nnn) to all the likely culprits.
>> Just be careful to disable core dumps if you do something like this on a
>> Unix-like system. We used to have a problem with a multitude of core
>> dumps for java processes filling up the disk of vmgump when surefire
>> killed forked processes.
> Just to be clear, I presume you are not referring to
> @Test(timeout=nnn) causing dumps?
Yes - at least I wouldn't know if @Test did cause core dumps.
> Also, I presume the same problem applies to Continuum itself killing builds.
True, as it was true for Gump itself. Likely applies to Jenkins and
others as well.
On second thought it has been the FreeBSD jail and not vmgump where this
happened as core dumps seem to be disabled on Ubuntu by default.
Stefan
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by sebb <se...@gmail.com>.
On 13 September 2011 10:18, Stefan Bodewig <bo...@apache.org> wrote:
> On 2011-09-13, sebb wrote:
>
>> In case it helps anyone else, I updated the pom as follows:
>
>> <artifactId>maven-surefire-plugin</artifactId>
>> <configuration>
>> <!-- Don't allow test to run for more than 30 minutes -->
>> <forkedProcessTimeoutInSeconds>1800</forkedProcessTimeoutInSeconds>
>
>> This kills the test; however the output only reports the test class,
>> not the individual method, and the surefire report file is empty.
>
>> To get better info, add @Test(timeout=nnn) to all the likely culprits.
>
> Just be careful to disable core dumps if you do something like this on a
> Unix-like system. We used to have a problem with a multitude of core
> dumps for java processes filling up the disk of vmgump when surefire
> killed forked processes.
Just to be clear, I presume you are not referring to
@Test(timeout=nnn) causing dumps?
AFAIK, that just kills the test thread, not the entire process.
Also, I presume the same problem applies to Continuum itself killing builds.
> Stefan
>
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by Stefan Bodewig <bo...@apache.org>.
On 2011-09-13, sebb wrote:
> In case it helps anyone else, I updated the pom as follows:
> <artifactId>maven-surefire-plugin</artifactId>
> <configuration>
> <!-- Don't allow test to run for more than 30 minutes -->
> <forkedProcessTimeoutInSeconds>1800</forkedProcessTimeoutInSeconds>
> This kills the test; however the output only reports the test class,
> not the individual method, and the surefire report file is empty.
> To get better info, add @Test(timeout=nnn) to all the likely culprits.
Just be careful to disable core dumps if you do something like this on a
Unix-like system. We used to have a problem with a multitude of core
dumps for java processes filling up the disk of vmgump when surefire
killed forked processes.
Stefan
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by sebb <se...@gmail.com>.
On 13 September 2011 00:21, Brett Porter <br...@apache.org> wrote:
>
> On 13/09/2011, at 3:27 AM, sebb wrote:
>
>>
>>>>
>>>> BTW, is there a way to limit the time Continuum allows tests to run?
>>>
>>> Done… but it will probably leave any dead builds behind.
>>
>> Is that a system-wide setting?
>
> It's a setting for the builds on that schedule. Since commons was on a separate schedule it hadn't been set, but the main one was.
What's the setting now please?
Continuum does not seem to allow even read-only access to such settings.
>>
>>> Do you have any idea what's causing them to hang?
>>
>> No idea why they are hanging yet, but I have raised the issue in Commons.
>> We can hopefully add a test timeout, which should avoid the zombies.
>
> Cool, thanks!
In case it helps anyone else, I updated the pom as follows:
<artifactId>maven-surefire-plugin</artifactId>
<configuration>
<!-- Don't allow test to run for more than 30 minutes -->
<forkedProcessTimeoutInSeconds>1800</forkedProcessTimeoutInSeconds>
This kills the test; however the output only reports the test class,
not the individual method, and the surefire report file is empty.
To get better info, add @Test(timeout=nnn) to all the likely culprits.
> - Brett
>
> --
> Brett Porter
> brett@apache.org
> http://brettporter.wordpress.com/
> http://au.linkedin.com/in/brettporter
>
>
>
>
>
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by Brett Porter <br...@apache.org>.
On 13/09/2011, at 3:27 AM, sebb wrote:
>
>>>
>>> BTW, is there a way to limit the time Continuum allows tests to run?
>>
>> Done… but it will probably leave any dead builds behind.
>
> Is that a system-wide setting?
It's a setting for the builds on that schedule. Since commons was on a separate schedule it hadn't been set, but the main one was.
>
>> Do you have any idea what's causing them to hang?
>
> No idea why they are hanging yet, but I have raised the issue in Commons.
> We can hopefully add a test timeout, which should avoid the zombies.
Cool, thanks!
- Brett
--
Brett Porter
brett@apache.org
http://brettporter.wordpress.com/
http://au.linkedin.com/in/brettporter
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by sebb <se...@gmail.com>.
On 12 September 2011 18:05, Brett Porter <br...@apache.org> wrote:
>
> On 13/09/2011, at 2:19 AM, sebb wrote:
>
>> Commons Pool has been running for over 20 hours:
>>
>> http://vmbuild.apache.org/continuum/buildResults.action?projectId=98&projectGroupId=16
>>
>> Please could someone kill the build?
>> Thanks.
>
> Done.
Thanks!
>>
>> BTW, is there a way to limit the time Continuum allows tests to run?
>
> Done… but it will probably leave any dead builds behind.
Is that a system-wide setting?
> Do you have any idea what's causing them to hang?
No idea why they are hanging yet, but I have raised the issue in Commons.
We can hopefully add a test timeout, which should avoid the zombies.
> - Brett
>
>
> --
> Brett Porter
> brett@apache.org
> http://brettporter.wordpress.com/
> http://au.linkedin.com/in/brettporter
>
>
>
>
>
Re: Continuum - Commons Pool unit test is stuck (again)
Posted by Brett Porter <br...@apache.org>.
On 13/09/2011, at 2:19 AM, sebb wrote:
> Commons Pool has been running for over 20 hours:
>
> http://vmbuild.apache.org/continuum/buildResults.action?projectId=98&projectGroupId=16
>
> Please could someone kill the build?
> Thanks.
Done.
>
> BTW, is there a way to limit the time Continuum allows tests to run?
Done… but it will probably leave any dead builds behind. Do you have any idea what's causing them to hang?
- Brett
--
Brett Porter
brett@apache.org
http://brettporter.wordpress.com/
http://au.linkedin.com/in/brettporter