In CentOS 6, puppet Unresponsive
Juan-Francisco Diez
We have the server installed in CentOS 5.7. The version installed of puppet-server is 2.7.1.
Some puppet clients have installed CentOS 6.x and the version 2.7.9 of puppet client.
The problem we have is that the puppet client become unresponsive after a few minutes. All we can see through the logs of the client is that the puppet is running but never end. We see the process through a "ps" command but it never finish. If we restart the service the node work properly for a few minutes.
We try updating the ruby and all of gems, the OS with the lastest packet but the problem still present. We don't know why it occurs. Also another nodes with the same packets and configuration work fine.
I hope someone can help with this trouble.
Happy new year to all of us.
Best Regards,
Juan-Francisco Diez Léglise
Denmat
--
You received this message because you are subscribed to the Google Groups "Puppet Users" group.
To post to this group, send email to puppet...@googlegroups.com.
To unsubscribe from this group, send email to puppet-users...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en.
Russell Van Tassell
Similarly, I believe the docs very clearly say (in one or more places) that your server always needs to be the newest version out of any of your clients (ie. The server is usually happy/fine to deal with older clients (with a few potential exceptions), but a newer client may not work with an older server).
--
Russell M. Van Tassell
russ...@gmail.com
This message was sent using my wireless handheld device.
Juan-Francisco Diez
I will try to put all in the same version
Russell Van Tassell
You can use RVM, or getting a ruby 1.8.7 build/rpm shouldn't be that difficult (I will have to look at our source, tomorrow, but we store a similar thing in our own local EL repo, pretty much for puppet and one or two other things ... At this hour I can't remember if we compiled it or grabbed it from a reputable upstream repository, however).
--
Russell M. Van Tassell
russ...@gmail.com
This message was sent using my wireless handheld device.
Dan White
Might be an easier solution.
“Sometimes I think the surest sign that intelligent life exists elsewhere in the universe is that none of it has tried to contact us.”
Bill Waterson (Calvin & Hobbes)
Craig White
On Dec 30, 2011, at 3:31 AM, Juan-Francisco Diez wrote:
> Thank to all. We try to put the server in the newest version. But with ruby in CentOS it's a bit more complicated because the new version is not packaged in the 5.x version.
----
I think you will find it much easier to remove any puppet packages installed on your CentOS 5.x server and just use the puppet gem instead
Craig
Craig White
On Dec 30, 2011, at 3:36 AM, Russell Van Tassell wrote:
> You can use RVM, or getting a ruby 1.8.7 build/rpm shouldn't be that difficult (I will have to look at our source, tomorrow, but we store a similar thing in our own local EL repo, pretty much for puppet and one or two other things ... At this hour I can't remember if we compiled it or grabbed it from a reputable upstream repository, however).
----
OP is talking about puppet version, you are talking about ruby version... two different things.
Craig
Craig White
On Dec 30, 2011, at 6:59 AM, Dan White wrote:
> Consider downgrading the CentOS 6 clients.
>
> Might be an easier solution.
----
probably not the best idea. his server is 2.7.1 and there's been a lot of bug fixes between 2.7.1 and 2.7.9 thus bringing his server up to speed makes more sense long term.
Craig
Dan White
Excellent point.
This will help me to remember keep track of server/client versions.
Dave Alden
On Dec 30, 2011, at 3:50 AM, Juan-Francisco Diez wrote:
> The problem we have is that the puppet client become unresponsive after a few minutes. All we can see through the logs of the client is that the puppet is running but never end. We see the process through a "ps" command but it never finish. If we restart the service the node work properly for a few minutes.
Just a "me too". Only I'm running 2.7.9 on a CentOS 6.2 VM and my clients are mostly 2.7.9 on Scientific Linux 6.1 (they haven't released 6.2 yet). I have a few CentOS 6.2 clients and I don't believe they've locked up - but many of my SL 6.1.
Unfortunately, restarting puppet "fixes" the problem for a random amount of time.
I haven't had time to dig into this any further - I was sorta hoping it was a RHEL/CentOS/SL 6.1 issue that is fixed in 6.2. :-)
...dave
Russell Van Tassell
Ummm... the thread changed a bit (and you cut out the pertinent part of the prior message). At least to me, it sounded like the OP was also questioning the Ruby version. From the message immediately prior to mine (which you conveniently snipped out - emphasis is mine):
-- begin
On Dec 30, 2011 2:31 AM, "Juan-Francisco Diez" <moc...@gmail.com> wrote:
Thank to all. We try to put the server in the newest version. But with ruby in CentOS it's a bit more complicated because the new version is not packaged in the 5.x version.
Nicolas Aizier
Juan-Francisco Diez
Got nearly 300 servers in RHEL 6.0, 6.1 and 6.2 all managed by puppet and 20 of them got that behavior.It's not related to the RHEL version neither the kernel version.Still digging on my side, if anyone find a solution please post.
--
You received this message because you are subscribed to the Google Groups "Puppet Users" group.
To view this discussion on the web visit https://groups.google.com/d/msg/puppet-users/-/QdENTILit9QJ.
