New systemd FYI & Question

[jwk]

Forgot to mention:
- Try installing another QM program like ORCA or MOPAC to see if the problem is g16-specific.

[jwk]

Also: If it is still not working, try installing ORCA or MOPAC to see if this is a QM program issue, or a WebMO issue.

[schmidt]

What does the output file say, if anything? By the time WebMO gets to that point, it will at least *try* to run Gaussian -- and should report any errors to the output file.

Note that Gaussian code requires any users to be members of the 'gaussian' group. This will likely require adding the 'webmo' user to that group as well.

[mosera]

What does the output file say, if anything? By the time WebMO gets to that point, it will at least *try* to run Gaussian -- and should report any errors to the output file.

Note that Gaussian code requires any users to be members of the 'gaussian' group. This will likely require adding the 'webmo' user to that group as well.

The user is a member of the gaussian group. I can run jobs as that user in its home directory and in /tmp with no issues. run_log ...

Executing script: /home/webmo/public_html/cgi-bin/webmo/run_gaussian.cgi
Creating working directory: /tmp/webmo-18644/11
Script execution node: [name of server correct]
Job execution node(s): [name of server correct]
Executing command: /usr/local/g16/g16

output.log

Entering Gaussian System, Link 0=/usr/local/g16/g16
Initial command:
/usr/local/g16/l1.exe "/tmp/webmo-18644/11/Gau-2609.inp" -scrdir="/tmp/webmo-18644/11/"

Its like it gets all the way to running, then just stops. It does wipe out the directory in the /tmp directory though when it fails. (FYI - I removed the extra hard drive with /scratch just to simplify the situation that used to hold scratch files.)

I was almost completely convinced that it was an selinux issue, but even if completely disabled selinux same issue (normally on permissive). I also checked /var/log/audit/audit.log and no issues there. Any thoughts on possible print statements I could add to run_gaussian.cgi to test why it is stopping after the run starts?

I'm also going to work on installing another software to check if that is an issue.

[doachs]

I think I ran into this same issue and found a way to fix it. In my case I am running on Rocky Linux 10.1. The issue is caused by the way the apache service is started on this OS. I needed to change the option MemoryDenyWriteExecute to no. The way I did that was to create a new conf file called /etc/systemd/system/httpd.service.d/nopt.conf and add this to it:

[Service]
PrivateTmp=false
ProtectHome=no
MemoryDenyWriteExecute=no

After I did that, reloaded the systemd stuff, and restarted apache, I no longer had failures running Gaussian.