[GE users] SGE 6.1 and qsh failure

Barry McInnes Barry.J.Mcinnes at noaa.gov
Mon Jul 23 14:19:19 BST 2007


    [ The following text is in the "iso-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]


>> To make qlogin and qrsh to work via ql.sh I have to swap the input
>> parameters around.
> 
> As stated in the Howto, you need this script only for qlogin. If you
> revert your setup to the Howto settings, it should work.
> 

Maybe its something in my setup but I have a echo logging ql.sh usage
and qrsh xterm generates an xterm, when I quit the xterm I get a log
from ql.sh, so something is calling it ?
The first line is the initial parameters passwd to ql.sh.
The second line is the parameters I pass to ssh on the last line

ql.sh Mon Jul 23 07:05:54 MDT 2007 -p 62148 mac56.cdc.noaa.gov exec
'/usr/local/sge/utilbin/darwin-ppc/qrsh_starter'
'/private/Network/Servers/g5s1/Volumes/XRAID/Users/sge/default/spool/mac56/active_jobs/34923.1'
pass  exec '/usr/local/sge/utilbin/darwin-ppc/qrsh_starter'
'/private/Network/Servers/g5s1/Volumes/XRAID/Users/sge/default/spool/mac56/active_jobs/34923.1'


from the job
SGE 6.1  Mon Jul 23 07:06:33 MDT 2007

Active Jobs
Where   JobID       User  Queue    Jobname     S  DD:HH:MM:SS     Mem MB
 mac56  34923   bmcinnes    low         xterm dr   0:00:00:00    0.00000


the job does not terminate correctly.
I get similar logs when just using qrsh, but with no exec parameters.

ql.sh Mon Jul 23 07:10:30 MDT 2007 -p 57239 mac36.cdc.noaa.gov
pass

> 
> No, don't use port 22. This will not use the SGE started sshd on the
> node, but the system sshd. This means of course to diable the firewall
> on the nodes.

Unfortunately we have to have the firewall on. Is there a specific range
I could open up ? From my logs it seems like 51000-63000 would be a
valid range.
I will try and open up the ports and see it that makes a difference....


> 
>> But qsh still gives the same error, and does not use ql.sh even though
>> my conf has
>> qlogin_command               /usr/local/sge/ql.sh
>> qlogin_daemon                /usr/sbin/sshd -i
>> rlogin_command               /usr/local/sge/ql.sh
>> rlogin_daemon                /usr/sbin/sshd -i
>> rsh_command                  /usr/local/sge/ql.sh
>> rsh_daemon                   /usr/sbin/sshd -i
>>
>> there must be another definition that qsh uses, ot it just fails to
>> schedule so it dosnt make it to the al.sh script ?
> 
> qsh will not use this settings at all. It will only start the command on
> the node, which is defined in the SGE configuration:
> 
> xterm                        /usr/bin/X11/xterm
> 
> Nothing is started on your local node. Hence this commmand will not work
> in your environment, as the xterm on the node will only try to get
> access to a port 6000+ on your local machine.
> 
> -- Reuti
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net
> 

-- 
---
Barry McInnes
325 Broadway
Boulder CO 80304
(303)4976231
barry.j.mcinnes at noaa.gov
---

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list