3.3.12 : podman xelatex requires executing a podman command to work

Administrators
1

Hi,

I upgraded to 3.3.12 and configured Podman to run xelatex, as explained in the documentation, after updating indico.cll and recompiling the SELinux module. I got the generic error about compilation failure:

indico.legacy.pdfinterface.latex.LaTeXRuntimeException: LaTeX compilation of /pdisk/indico/tmp/indico-texgen-oao75d6m/report.tex failed

After double checking everything, I realized that to get PDF generation to work, I had to execute once a podman command (no matter which one, for example podman ps) under the indico account after a (virtual) machine reboot to fix the problem.

I don’t figure out what could be the reason. Did somebody face a similar issue? It"s working with this (easy) workaround but I’ve the feeling it is hiding something misconfigured and I’d like to identify what!

Cheers,

2

Does /pdisk/indico/tmp/indico-texgen-oao75d6m still exist? If yes, check the log files in there. It contains more details that may be useful.

3

I forgot to mention that this directory does exist but the output.log is empty…

Checking once again our “historical” configuration, I realized that we create the indico account with a group also called indico rather than apache (or nginx when using Nginx). I had in the past to set the group of some directories (logs, attachements) to apache but apart from that I never spotted any error because of that. I don’t think it has a real impact here but just want to mention it in case it could explain the misbehaviour…

4

Does the selinux audit.log contain anything interesting for when running podman failed?

5

No SELinux doesn’t report any error (and disabling it doesn’t help) as long as the SELinux module is the updated one provided for 3.3.12.

6

BTW, I use a recent YUM snapshot (from May 4) and the Podman version is 5.6.0.

7

ok, weird. anyway since running podman fixed it for you I don’t think it’s worth spending much time on trying to find out the original cause…

8

Agreeed! I remain interested if someone experiences the same issue as I don’t see what we do which is not very standard…

9

I have similar effect when running podman in docker. After a reboot podman complains about a wrong boot id (I still need to configure the tmpdir correctly).

That seems to be a known problem if the podman storage is not in a tmpfs (see Weird error message after reboot: Error: current system boot ID differs from cached boot ID · podman-container-tools/podman · Discussion #23193 · GitHub ). If in your vm the podman tmp dir did not get cleaned correctly, I would expect a similar effect.