WebMO 17 Install

[layfield]

Is there an archive of installation instructions? I have an Enterprise version of WebMO 17.

I am trying to re-install WebMO on a Rocks 6.2 cluster (based on CentOS 6.6). When i follow the Apache (v. 2.2.15) instructions, I get an error saying that there is an internal server error.

I am not sure if these are related issues but I am trying to chase down all possible issues.

Thanks!

[polik]

There is no difference in the installation procedure for WebMO on Linux between versions 17 and 20.

For completeness, however, the version 17 instructions are attached.

[layfield]

Thank you for the instructions. Is it possible that the Apache settings have changed between versions?

When i remove the line "Require method GET POST OPTIONS PUT DELETE PATCH' from my httpd.conf file, I am able to get the site to load and it appears that i can get full functionality when i run on the headnode. I am unable to connect to the SGE queueing system.

I thank you for your consideration as this is likely not a WebMO issue but I appreciate your answer.

Thank you!

[schmidt]

You should first ensure that you can access SGE and submit jobs from the command line as the user 'webmo'. If that fails, you would need to consult the SGE docs. Once that is working I can assist with WebMO integration.

[layfield]

Thank you for your offer of help.

1) I am able to access qsub (and all other sge commands) from the command line.

2) When I try to setup my Apache server as described in the online instructions, I get an internal server error and the following error message in /var/log/httpd/error_log:
configuration error: couldn't perform authentication. AuthType not set!: /~webmo/cgi-bin/webmo/login.cgi

I have tried to google the error and the only (partially helpful solution that I found was to remove the "Require" statement as I suggested previously. When I do this I am able to setup WebMO on the headnode with full functionality. When I try to interface with SGE (which works fine for job submission from the commandline), I get the following error:

error: commlib error: got select error (Connection refused)
Unable to run job: unable to send message to qmaster using port 6444 on host "raven.local": got send error.
Exiting.


If you have any additional suggestions they would be greatly appreciated.

Thank you!

[GGHoffman]

I have the same problem. I updated to WebMO Enterprise 20 from WebMO Enterprise 17. Any job I submit now returns the message:

error: commlib error: got select error (Connection refused)
Unable to run job: unable to send message to qmaster using port 6444 on host "parcore.local": got send error.
Exiting.

I can submit jobs as webmo - or any other user - directly from the terminal using the command "qsub ... ". The only change made to the system is the upgrade to WebMO. I have never had a problem upgrading WebMO before. Can you suggest something I can do to resolve this? Thanks,

Gary

[schmidt]

Make sure that the the variable 'sge_qmaster' is set/defined to the appropriate port in /etc/services. This is often set by environmental variable during login (you can check via 'printenv'), but WebMO cannot see that information! As such, it needs to get set in /etc/services.

[GGHoffman]

Thanks for your reply. It took a little more digging, but I was able to come up with a solution. Perhaps others with the same problem may benefit. My system uses the Rocks operating system (v. 6) and I am using the SGE queuing system. I ran "ss -p | grep qmaster" and viewed the port numbers for the associated sockets. All the compute nodes used port number 536 - not 6444 as is listed for sge_qmaster in /etc/services! I don't know if this a quirk of the operating system, but that's what was there. So, in processcontrol_pbs.cgi, after the line "if ($externalBatchQueue eq 'sge')", I added the line "$sgeQmasterPort=536;" and everything now seems to work. If this is something that will cause other problems, please let me know. Thanks for your help.