Hi Felix,
Since I wrote this for somebody else on a different mailing list yesterday:
You'll need to create a munge key somewhere with
'/usr/sbin/create-munge-key', copy it to '/etc/munge/munge.key' on all the
nodes then start the munge service everywhere as well.
Presuming you are using the same version of torque everywhere that should
allow the nodes to talk to each other.
You may also need to add 'set server authorized_users += *@<hostname>' to
the toprque server config for all your CEs.
What we've done to prevent surprises like this is to create yum
repositories
for torque and maui and exclude them from gLite/UMD/EMI/EPEL (we're looking
at using yum priorities) then we manually add new versions at the point
_we_
wish to update.
Yours,
Chris Brew.
On 03/11/2011 07:47, "felix farcas" <[log in to unmask]> wrote:
>Hello
>
>Accidentally I did an upgrade on all my worker nodes to torque 5.x
>
>My
>
>I got an interesting error when i try to launch qstat -q:
>
> munge: Error: Unable to access "/var/run/munge/munge.socket.2": No
>such file or directory
>qstat: End of File (null)
>
>How may I overcome this error?
>
>Thank you
>Felix
>--
>Dr. Ing. Farcas Felix
>National Institute of Research and Development
>of Isotopic and Molecular Technology,
>IT - Department - Cluj-Napoca, Romania
>
|