Surefire Jobs

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

Surefire Jobs

Karl Heinz Marbaise-3
Hi,

it seemed to me that there are Surefire Jobs stuck ..for 10 hours or more...

I would like to abort them ?

Kind regards
Karl Heinz Marbaise

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Surefire Jobs

Karl Heinz Marbaise-3
Hi,

On 18/02/18 14:51, Karl Heinz Marbaise wrote:
> Hi,
>
> it seemed to me that there are Surefire Jobs stuck ..for 10 hours or
> more...
>
> I would like to abort them ?

Seemed to be gone already...

Kind regards
Karl Heinz Marbaise
>
> Kind regards
> Karl Heinz Marbaise
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Surefire Jobs

Karl Heinz Marbaise-3
Hi Tibor,

On 18/02/18 15:13, Tibor Digana wrote:
> no, no
> Please do not do it!

I would have at least waited for a feedback before I kill a Job ;-)...


> We are waiting for them.
> They are expected to run 12 hours.

Wow that long..I wasn't aware of that...

> You can compute it:
> 1 hour each run configuration.
> Number of configuration is 3 Maven config * 4 JDKs.


>
> I tried to run them in parallel but they appeared several in one executor.
> Maybe later as parallel but now before release it is not safe. We will
> rather wait longer.

Of course not...But I think we should take a deeper look at them and see
if we can get them faster ...after the release...


Kind regards
Karl Heinz Marbaise

>
> On Sun, Feb 18, 2018 at 2:56 PM, Karl Heinz Marbaise <[hidden email]>
> wrote:
>
>> Hi,
>>
>> On 18/02/18 14:51, Karl Heinz Marbaise wrote:
>>
>>> Hi,
>>>
>>> it seemed to me that there are Surefire Jobs stuck ..for 10 hours or
>>> more...
>>>
>>> I would like to abort them ?
>>>
>>
>> Seemed to be gone already...
>>
>>
>> Kind regards
>> Karl Heinz Marbaise
>>
>>>
>>> Kind regards
>>> Karl Heinz Marbaise

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: Surefire Jobs

Tibor Digana
No reason.
One job always took 1 hour. You do not have to.
We are working on a release. If you commit anything to master I can start
from the begin.
After the release I will use parallel execution but I could not because
Gavin from our INFRA team installed latest Pipeline Advanced Utilities few
hour ago.
There I needed lock() and zip().
Why lock?
Because failsafe-plugin has 4 integration tests and they bind to port 8083.
If I run them in parallel, one cannot bind.
I could find random port via helper-plugin but this was not guarantee
either.


On Sun, Feb 18, 2018 at 3:18 PM, Karl Heinz Marbaise <[hidden email]>
wrote:

> Hi Tibor,
>
> On 18/02/18 15:13, Tibor Digana wrote:
>
>> no, no
>> Please do not do it!
>>
>
> I would have at least waited for a feedback before I kill a Job ;-)...
>
>
> We are waiting for them.
>> They are expected to run 12 hours.
>>
>
> Wow that long..I wasn't aware of that...
>
> You can compute it:
>> 1 hour each run configuration.
>> Number of configuration is 3 Maven config * 4 JDKs.
>>
>
>
>
>> I tried to run them in parallel but they appeared several in one executor.
>> Maybe later as parallel but now before release it is not safe. We will
>> rather wait longer.
>>
>
> Of course not...But I think we should take a deeper look at them and see
> if we can get them faster ...after the release...
>
>
>
> Kind regards
> Karl Heinz Marbaise
>
>
>> On Sun, Feb 18, 2018 at 2:56 PM, Karl Heinz Marbaise <[hidden email]>
>> wrote:
>>
>> Hi,
>>>
>>> On 18/02/18 14:51, Karl Heinz Marbaise wrote:
>>>
>>> Hi,
>>>>
>>>> it seemed to me that there are Surefire Jobs stuck ..for 10 hours or
>>>> more...
>>>>
>>>> I would like to abort them ?
>>>>
>>>>
>>> Seemed to be gone already...
>>>
>>>
>>> Kind regards
>>> Karl Heinz Marbaise
>>>
>>>
>>>> Kind regards
>>>> Karl Heinz Marbaise
>>>>
>>>
Reply | Threaded
Open this post in threaded view
|

Re: Surefire Jobs

olamy
what is the issue with helper-plugin? It's supposed to work.
can you explain?

I would prefer having this fix in the maven build itself rather than
depends on some Jenkins extra configuration (make it easier for people to
test in their own ci instance)
perso I find this build too long in a CI perspective. You have to wait very
long to get feedback. Maybe reducing the number of maven version tested?

On 19 February 2018 at 00:22, Tibor Digana <[hidden email]> wrote:

> No reason.
> One job always took 1 hour. You do not have to.
> We are working on a release. If you commit anything to master I can start
> from the begin.
> After the release I will use parallel execution but I could not because
> Gavin from our INFRA team installed latest Pipeline Advanced Utilities few
> hour ago.
> There I needed lock() and zip().
> Why lock?
> Because failsafe-plugin has 4 integration tests and they bind to port 8083.
> If I run them in parallel, one cannot bind.
> I could find random port via helper-plugin but this was not guarantee
> either.
>
>
> On Sun, Feb 18, 2018 at 3:18 PM, Karl Heinz Marbaise <[hidden email]>
> wrote:
>
> > Hi Tibor,
> >
> > On 18/02/18 15:13, Tibor Digana wrote:
> >
> >> no, no
> >> Please do not do it!
> >>
> >
> > I would have at least waited for a feedback before I kill a Job ;-)...
> >
> >
> > We are waiting for them.
> >> They are expected to run 12 hours.
> >>
> >
> > Wow that long..I wasn't aware of that...
> >
> > You can compute it:
> >> 1 hour each run configuration.
> >> Number of configuration is 3 Maven config * 4 JDKs.
> >>
> >
> >
> >
> >> I tried to run them in parallel but they appeared several in one
> executor.
> >> Maybe later as parallel but now before release it is not safe. We will
> >> rather wait longer.
> >>
> >
> > Of course not...But I think we should take a deeper look at them and see
> > if we can get them faster ...after the release...
> >
> >
> >
> > Kind regards
> > Karl Heinz Marbaise
> >
> >
> >> On Sun, Feb 18, 2018 at 2:56 PM, Karl Heinz Marbaise <[hidden email]
> >
> >> wrote:
> >>
> >> Hi,
> >>>
> >>> On 18/02/18 14:51, Karl Heinz Marbaise wrote:
> >>>
> >>> Hi,
> >>>>
> >>>> it seemed to me that there are Surefire Jobs stuck ..for 10 hours or
> >>>> more...
> >>>>
> >>>> I would like to abort them ?
> >>>>
> >>>>
> >>> Seemed to be gone already...
> >>>
> >>>
> >>> Kind regards
> >>> Karl Heinz Marbaise
> >>>
> >>>
> >>>> Kind regards
> >>>> Karl Heinz Marbaise
> >>>>
> >>>
>



--
Olivier Lamy
http://twitter.com/olamy | http://linkedin.com/in/olamy
Reply | Threaded
Open this post in threaded view
|

Re: Surefire Jobs

Tibor Digana
I am using the lock() in my company. The helper-plugin finds a port but two
jobs on one machine are so predicable that this mechanism would not be
reliable.
It happened to me in my company. The helper may help but it does not
guarantee anything.

There are tests which work but they are not much reliable if e.g. the
machine is overloaded by neighboring job on the same machine.
For instance ParallelComputerBuilderTest computes exec time of parallel
test which can be perfectly computed. The delays are 0.5 second and they
should be longer. It's obvious why.
The next is the Integration Test 1295 which has some race condition. See
another e-mail in our mailing list. I am discussing it already with the
person who created the IT and the native library too.

It does not make sense to have parallel CI jobs unless I will improve these
tests.
The point is first to understand the internals of the project from the
history and the other things will come by themselves.
After the improvements are made the parallel execution can be used and the
likelihood that one test breaks due to concurrency is lower.
Notice we have 1400 tests altogether and 21 Jenkins jobs and none of the
tests must fail.

We should make a deal.
We should not directly commit to the master unless we make a code review
together in a particular branch.
The branch should be open with name e.g. SUREFIRE-12345 which would be the
name of Jira issue.
I use to have one commit per branch (wasn't last time I know) and I use *git
--amend* till it's acceptable.
The commit has name "[SUREFIRE-12345] Title in JIRA".

It's too late.
Let's talk tomorrow evening.

Cheers
Tibor




On Sun, Feb 18, 2018 at 11:26 PM, Olivier Lamy <[hidden email]> wrote:

> what is the issue with helper-plugin? It's supposed to work.
> can you explain?
>
> I would prefer having this fix in the maven build itself rather than
> depends on some Jenkins extra configuration (make it easier for people to
> test in their own ci instance)
> perso I find this build too long in a CI perspective. You have to wait very
> long to get feedback. Maybe reducing the number of maven version tested?
>
> On 19 February 2018 at 00:22, Tibor Digana <[hidden email]> wrote:
>
> > No reason.
> > One job always took 1 hour. You do not have to.
> > We are working on a release. If you commit anything to master I can start
> > from the begin.
> > After the release I will use parallel execution but I could not because
> > Gavin from our INFRA team installed latest Pipeline Advanced Utilities
> few
> > hour ago.
> > There I needed lock() and zip().
> > Why lock?
> > Because failsafe-plugin has 4 integration tests and they bind to port
> 8083.
> > If I run them in parallel, one cannot bind.
> > I could find random port via helper-plugin but this was not guarantee
> > either.
> >
> >
> > On Sun, Feb 18, 2018 at 3:18 PM, Karl Heinz Marbaise <[hidden email]>
> > wrote:
> >
> > > Hi Tibor,
> > >
> > > On 18/02/18 15:13, Tibor Digana wrote:
> > >
> > >> no, no
> > >> Please do not do it!
> > >>
> > >
> > > I would have at least waited for a feedback before I kill a Job ;-)...
> > >
> > >
> > > We are waiting for them.
> > >> They are expected to run 12 hours.
> > >>
> > >
> > > Wow that long..I wasn't aware of that...
> > >
> > > You can compute it:
> > >> 1 hour each run configuration.
> > >> Number of configuration is 3 Maven config * 4 JDKs.
> > >>
> > >
> > >
> > >
> > >> I tried to run them in parallel but they appeared several in one
> > executor.
> > >> Maybe later as parallel but now before release it is not safe. We will
> > >> rather wait longer.
> > >>
> > >
> > > Of course not...But I think we should take a deeper look at them and
> see
> > > if we can get them faster ...after the release...
> > >
> > >
> > >
> > > Kind regards
> > > Karl Heinz Marbaise
> > >
> > >
> > >> On Sun, Feb 18, 2018 at 2:56 PM, Karl Heinz Marbaise <
> [hidden email]
> > >
> > >> wrote:
> > >>
> > >> Hi,
> > >>>
> > >>> On 18/02/18 14:51, Karl Heinz Marbaise wrote:
> > >>>
> > >>> Hi,
> > >>>>
> > >>>> it seemed to me that there are Surefire Jobs stuck ..for 10 hours or
> > >>>> more...
> > >>>>
> > >>>> I would like to abort them ?
> > >>>>
> > >>>>
> > >>> Seemed to be gone already...
> > >>>
> > >>>
> > >>> Kind regards
> > >>> Karl Heinz Marbaise
> > >>>
> > >>>
> > >>>> Kind regards
> > >>>> Karl Heinz Marbaise
> > >>>>
> > >>>
> >
>
>
>
> --
> Olivier Lamy
> http://twitter.com/olamy | http://linkedin.com/in/olamy
>