Re: precommit zombies

2013-01-14 Thread Nicolas Liochon
:-) I've added a kill in the scripts. Before and after our build. I wanted to look around a little before doing so, but without the ssh access, I've chosen a more definitive solution :-) On Mon, Jan 14, 2013 at 6:44 PM, Todd Lipcon wrote: > Yep, I have access if anything needs to get installed,

Re: precommit zombies

2013-01-14 Thread Todd Lipcon
Yep, I have access if anything needs to get installed, etc. But probably best to figure out anti-zombie provisions that don't involve me SSHing on a regular basis to kill processes :) On Mon, Jan 14, 2013 at 9:17 AM, Stack wrote: > On Mon, Jan 14, 2013 at 8:59 AM, Nicolas Liochon wrote: > >> On

Re: precommit zombies

2013-01-14 Thread Stack
On Mon, Jan 14, 2013 at 8:59 AM, Nicolas Liochon wrote: > On Jenkins, when we're admin, it does not mean that we can log in the build > machine, yes? > I tried to connect by different means, including from people.apache.org, > but it failed... > > That is right. I asked for it once and was laugh

Re: precommit zombies

2013-01-14 Thread Nicolas Liochon
On Jenkins, when we're admin, it does not mean that we can log in the build machine, yes? I tried to connect by different means, including from people.apache.org, but it failed... On Wed, Jan 9, 2013 at 5:20 PM, Nicolas Liochon wrote: > >We can run a few commands before the build starts (we do

Re: precommit zombies

2013-01-09 Thread Nicolas Liochon
>We can run a few commands before the build starts (we do already). Could add more reporting context test is running in. Yes, If it happens again I will. On Wed, Jan 9, 2013 at 5:12 PM, Stack wrote: > On Wed, Jan 9, 2013 at 8:08 AM, Nicolas Liochon wrote: > > > > You have admin privs on jenk

Re: precommit zombies

2013-01-09 Thread Stack
On Wed, Jan 9, 2013 at 8:08 AM, Nicolas Liochon wrote: > > You have admin privs on jenkins Nicolas? I can add. > Yes, please. It could be useful sometimes may be. > > It looks like you can just use your apache credentials and you should be in. > > Selective kill seems like good idea if jenkins

Re: precommit zombies

2013-01-09 Thread Nicolas Liochon
> You think these mess up our tests at all N? Theoritically, no. But if they consume memory & CPU it at the very least does not help us. > Or are these just the zombies sometimes reported in our test output on jenkins? Yes, it gives a wrong warning on our builds. May be it's just an exception, I d

Re: precommit zombies

2013-01-09 Thread Stack
On Wed, Jan 9, 2013 at 7:17 AM, Nicolas Liochon wrote: > FYI, We sometimes have remains of other product builds. You think these mess up our tests at all N? Or are these just the zombies sometimes reported in our test output on jenkins? > Lastly, HDFS, > likely HDFS-4261

Re: precommit zombies

2013-01-09 Thread Ted Yu
Looks like HDFS-4261 would be resolved today. We can keep an eye on non-HBase test timeout and inform upstream projects. Cheers On Wed, Jan 9, 2013 at 7:17 AM, Nicolas Liochon wrote: > FYI, We sometimes have remains of other product builds. Las