[torqueusers] No contact with server at hostaddr problem (followup)

nathaniel.x.woody at gsk.com nathaniel.x.woody at gsk.com
Tue Jul 10 14:20:24 MDT 2007


Tim,

I have managed to mis-configure pbs to give me these symptoms in two ways:

1) pbs_server isn't running on the port that the mom thinks it is.  Make 
sure that the pbs_server is running on 15001 (looks like you're already 
looking at this).  As mentioned not long ago, you can start the mom with 
pbs_mom -S 15000 to force mom to look for the server at port 15000. 
(though I suppose a wise thing to do may be to see what port pbs_server is 
running on first). 

2) Fudged up ethernet names for the server on the mom (personally, I've 
done this with multi-homed servers).  Does the mom-node (cree?) know who 
huron(server) is (and vice versa)?  Is that an entry in /etc/hosts on the 
mom-node?  Being a wimp, I almost always use the ip of the server in the 
config file for the pbsserver entry to avoid making this mistake.

I suppose the third option is that pbs_server is actually running at all 
on huron. 

Hope that helps,
Nate





"Carbo, Timothy J." <TIMOTHY.J.CARBO at saic.com> 
Sent by: torqueusers-bounces at supercluster.org
10-Jul-2007 15:16
 
To
"Garrick Staples" <garrick at usc.edu>, torqueusers at supercluster.org
cc

Subject
RE: [torqueusers] No contact with server at hostaddr problem (followup)






Garrick:

Sorry I wasn't clear

My set up is 

Node1 (cree):  running pbs_server, pbs_mom and maui

server_priv/nodes:
cree np=8
Huron np=8

mom_priv/config:
$pbsserver cree

Node2 (huron):  running pbs_mom only

mom_priv/config:
$pbsserver cree

When I submit the following on cree

echo "sleep 30" | qsub

the job appears to be scheduled on huron and runs OK but then I start
seeing the "No contact with server at hostaddr port 15001" error
messages repeated in the mom_logs file on huron and it appears that the
pbs_server never is notified that the job ran to completion.

Hope this clears things up a little.

Regards,
Tim


-----Original Message-----
From: torqueusers-bounces at supercluster.org
[mailto:torqueusers-bounces at supercluster.org] On Behalf Of Garrick
Staples
Sent: Tuesday, July 10, 2007 12:28 PM
To: torqueusers at supercluster.org
Subject: Re: [torqueusers] No contact with server at hostaddr problem
(followup)

On Mon, Jul 09, 2007 at 09:30:09AM -0600, Carbo, Timothy J. alleged:
> Hello all.
> 
> 
> 
> I was tracking the following email chain and was wondering if there is
> any resolution to the problem below.  I just installed TORQUE 2.1.8
with
> Maui 3.2.6-p19 on a two node system (both x86-64 bit Xeon quad core
> systems running Red Hat AS 4 update 4) and am having the same exact
> problem when I try to submit a job on my client node (jobs run fine on
> the server node).  Oddly, the remote node is trying to connect to port
> 15001 on the server node but netstat -a indicates there is nothing
> listening at that port.  I am pretty new to Torque so am I missing
> something?

It is a little hard to figure out your setup here with "client",
"server", and "remote" nodes.

If both hosts are to handle compute jobs, then you want pbs_mom running
on both hosts and both hostnames in server_priv/nodes.

-- 
Garrick Staples, GNU/Linux HPCC SysAdmin
University of Southern California

Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html
_______________________________________________
torqueusers mailing list
torqueusers at supercluster.org
http://www.supercluster.org/mailman/listinfo/torqueusers


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.supercluster.org/pipermail/torqueusers/attachments/20070710/979f6731/attachment-0001.html


More information about the torqueusers mailing list