At present it seems there is a 5 second timeout being enforced somewhere,
from reading those JIRA at least.... I'd like to spend some more time
investigating, but I have a new deadline looming for more enterprise
features to carve out!!! and I have to do some more stuff on literate, etc.
Hopefully something will shake out in the wash with the enterprise
ssh-slaves plugin. Certainly if I find any logic or core bugs I will push
them back, but unless PM decides otherwise the Non-blocking implementation
is for pay customers only :-(


On 7 October 2013 15:21, Henri Gomez <[email protected]> wrote:

> >[Shameless plug]
> >http://developer-blog.cloudbees.com/2013/10/ssh-slaves-reloaded.html
> >[/Shameless plug]
>
> Enterprise edition only isn't it ? ;(
>
>
>
> > Anyway, in 1.3 vs 1.2 I didn't change the ssh connection logic,
> similarly there is no changes outside of credentials since 0.27.
> >We'd really need to know more about the ssh server that is being
> connected to... also it may be a keep-alive failure... not sure how much I
> can improve that with the trilead impl.
>
> Problems occurs with CentOS 6.4 (openssh-server-5.3p1-84.1.el6.x86_64) and
> openSUSE 12.2 (openssh-6.0p1-2.3.3.x86_64), both in 64bits.
>
> It shouldn't be related to keep-alive as problem happen also by the end of
> build, so there is still activity.
>
> On slave side, I noticed when probem occurs
>
> Oct  7 01:30:53 swf-slave-1-opensuse-64 systemd-logind[539]: Removed
> session 652.
> Oct  7 01:31:03 swf-slave-1-opensuse-64 sshd[20742]: Accepted publickey
> for cijenka from 10.128.59.11 port 35785 ssh2
> Oct  7 01:31:03 swf-slave-1-opensuse-64 systemd-logind[539]: New session
> 1189 of user cijenka.
> Oct  7 01:31:03 swf-slave-1-opensuse-64 sshd[20747]: subsystem request for
> sftp by user cijenka
>
> A normal behavior is :
>
> Oct  7 09:20:12 swf-slave-1-opensuse-64 sshd[20747]: subsystem request for
> sftp by user cijenka
> Oct  7 09:20:13 swf-slave-1-opensuse-64 sshd[20747]: Received disconnect
> from 10.128.59.11: 11: Closed due to user request.
> Oct  7 09:20:13 swf-slave-1-opensuse-64 systemd-logind[539]: Removed
> session 1189.
> Oct  7 09:20:14 swf-slave-1-opensuse-64 sshd[14817]: Accepted publickey
> for cijenka from 10.128.59.11 port 40125 ssh2
> Oct  7 09:20:14 swf-slave-1-opensuse-64 systemd-logind[539]: New session
> 1277 of user cijenka.
> Oct  7 09:20:14 swf-slave-1-opensuse-64 sshd[14821]: subsystem request for
> sftp by user cijenka
>
> When problem appears, there is no "Closed due to user request" phase.
>
> Only change we notice in our configuration :
>
> * EnvInject is used by these jobs
> * We upgraded to Jenkins 1.529
>
> And we're not alone :
>
> * https://issues.jenkins-ci.org/browse/JENKINS-18781
> * https://issues.jenkins-ci.org/browse/JENKINS-18879
>
> I notice something weird for some time, still related to slave :
>
> if I try to get slave log, ie http://myjenkins/computer/myslavelog,
> nothing appears except a spinning wheel on top ;(
>
>
>
>
>
>  --
> You received this message because you are subscribed to the Google Groups
> "Jenkins Developers" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> For more options, visit https://groups.google.com/groups/opt_out.
>

-- 
You received this message because you are subscribed to the Google Groups 
"Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to