Jeremy
At 11:57 AM 3/2/2004, Benjamin Simmons wrote:
I added viper.memphis.edu and borg.memphis.edu to both the clienthosts and to the restricted parts and added the /home entries as well although both systems share a common home directory system.
The same error message persists.
Ben
Frank Crawford wrote:
Ben, Have a look in /var/spool/pbs/mon_priv at the file config. You may need to add either the restricted or clienthosts line for them. See pbs_mom man page for more details.
Frank
On Tue, 2004-03-02 at 12:14, Benjamin Simmons wrote:
I added +viper.memphis.edu to the etc/hosts.equiv on borg.memphis.edu. Viper is the machine that I want to do my job submission from, and borg.memphis.edu is the cluster server.
This is the error message listed in the logs on borg.memphis.edu
03/01/2004 19:09:42;0100;PBS_Server;Req;;Type 49 request received from [EMAIL PROTECTED], sock=11
03/01/2004 19:09:42;0080;PBS_Server;Req;req_reject;Reject reply code=15008, aux=0, type=49, from [EMAIL PROTECTED]
this is the error message I recieved when I tried the qsub on viper:
[EMAIL PROTECTED] bdsimmns]$ qsub -q [EMAIL PROTECTED] surface.pbs pbs_iff: error returned: 15008 pbs_iff: Access from host not allowed, or unknown host No Permission. qsub: cannot connect to server borg.memphis.edu (errno=15007) [EMAIL PROTECTED] bdsimmns]$
I can ssh to and from the two machines without a password, and they share a common home directory system. I can submit jobs to the queue workq while logged into borg.
Any other thoughts as to things to look for?
Ben
Frank Crawford wrote:
Ben, Did you add something in /etc/hosts.equiv on the other server (borg.memphis.edu)? Anyway, I think that problems with hosts.equiv come up with a 10523 (Bad User) error.
What are you seeing in the logs on the other server, i.e. borg.memphis.edu? You should either see something logged about the connection and possibly a rejection, or you would have a deeper level problem.
Frank
On Tue, 2004-03-02 at 10:41, Benjamin Simmons wrote:
ok, here is what I have setup right how:
I edited /var/spool/pbs/server_name and set it to the cluster server
I edited hosts.equiv and set it to +outside.computer.name
This computer is connected through eth1 of the cluster server, so I edited the pfilter.conf to have eth1 all trusted, to ensure that would not play a role for now.
I created the following script and executed it with qsub.
The error message follows below, but is permissions based. Thanks, Ben
[EMAIL PROTECTED] bdsimmns]$ vi surface.pbs
#PBS -N se_ #PBS -o AxiCircleNoDowel.log #PBS -e AxiCircleNoDowel.pbs cd /home/1/bdsimmns echo /home/1/bdsimmns/AxiCircleNoDowel.fe /home/1/bdsimmns/bin/evolver /home/1/bdsimmns/AxiCircleNoDowel.fe #All done ~
[EMAIL PROTECTED] bdsimmns]$ qsub -q [EMAIL PROTECTED] surface.pbs pbs_iff: error returned: 15008 pbs_iff: Access from host not allowed, or unknown host No Permission. qsub: cannot connect to server borg.memphis.edu (errno=15007) [EMAIL PROTECTED] bdsimmns]$
Jeremy Enos wrote:
The extra pbs server and routing queue shouldn't be necessary. Try setting up a hosts.equiv on the pbs_server machine which includes the machine you want to run qsub from. Also, on the machine you're running qsub from, set the server_name file appropriately for the remote server. qsub/qstat commands should work then. As I mentioned before though, I'm not positive that the hosts.equiv is all that is necessary, but I am sure that you don't need two pbs_servers.
Jeremy
At 03:03 PM 3/1/2004, Benjamin Simmons wrote:
I already went through that pain to get a username and password there. The way I read through the manuals is that I need a pbs_server running on the machine that I am making the submission from, and that I have a queue defined on this machine as well. This machine's queue is a routing queue, and I can define one or more destination queues. I for now only want it to go to the queue that is on my cluster server, but I will later need to direct it to other clusters around our campus.
I can send you privately the pdf of the admin guide from the site you mentioned, or reference pages I think I am using correctly.
Am I misunderstanding how this is supposed to work to go between different physical machines?
Thanks, Ben
Jeremy Enos wrote:
At 01:32 PM 3/1/2004, Benjamin Simmons wrote:
Ok, I think I set everything up correctly, but when I try to submit a job I get a error that job has been rejected by all possible destinations.Not sure why you would have server/queue configs at all on the machine you're submitting on... shouldn't have a server on there at all, right? I know there are admin guides available at http://www.openpbs.org that may also be of help. You will need to register a user name though. (I had some trouble finding a link to do this there though)
Any thoughts on what I need to look at, or do I need to post the server and queue configs for the machine I am submitting on and the one that is recieving?
Jeremy
Thanks for the help, Ben Simmons
Jeremy Enos wrote:
It's been awhile since I looked into it, but I think a hosts.equiv file is needed to allow submission from other hosts. I don't have reliable detail past that at the moment.
Jeremy
At 07:57 PM 2/26/2004, Benjamin Simmons wrote:
Has anyone tried to setup a pbs queue that routes to a different server outside their cluster? Or are most people just having users submit the job on the server that it should be run on?
I am getting many different types of errors, but am looking to see if there is an experience base to draw on here, or if I need to look towards another list,
Thanks to all in advance,
Ben Simmons The University of Memphis http://viper.memphis.edu
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id=1356&alloc_id=3438&op=click _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
