Summary: | pam_slurm_adopt and XDG variable | ||
---|---|---|---|
Product: | Slurm | Reporter: | Marco Induni <marco.induni> |
Component: | Other | Assignee: | Felip Moll <felip.moll> |
Status: | RESOLVED DUPLICATE | QA Contact: | |
Severity: | 4 - Minor Issue | ||
Priority: | --- | CC: | bart, felip.moll |
Version: | 17.11.12 | ||
Hardware: | Linux | ||
OS: | Linux | ||
See Also: | https://bugs.schedmd.com/show_bug.cgi?id=5920 | ||
Site: | CSCS - Swiss National Supercomputing Centre | Alineos Sites: | --- |
Atos/Eviden Sites: | --- | Confidential Site: | --- |
Coreweave sites: | --- | Cray Sites: | --- |
DS9 clusters: | --- | Google sites: | --- |
HPCnow Sites: | --- | HPE Sites: | --- |
IBM Sites: | --- | NOAA SIte: | --- |
NoveTech Sites: | --- | Nvidia HWinf-CS Sites: | --- |
OCF Sites: | --- | Recursion Pharma Sites: | --- |
SFW Sites: | --- | SNIC sites: | --- |
Tzag Elita Sites: | --- | Linux Distro: | --- |
Machine Name: | CLE Version: | ||
Version Fixed: | Target Release: | --- | |
DevPrio: | --- | Emory-Cloud Sites: | --- |
Description
Marco Induni
2018-11-12 09:04:52 MST
(In reply to Marco Induni from comment #0) > Dear support, > I followed the instructions (https://slurm.schedmd.com/pam_slurm_adopt.html) > to limit the access on the compute nodes where there is a JOB running and it > works. > I have only a problem discovered, because some environment variables now > aren't set anymore. > Indeed based on the instruction you should comment out in the > /etc/pam.d/system-auth the following > > #account sufficient pam_localuser.so > #-session optional pam_systemd.so > > The problem is that pam_systemd set for example all the XDG variables (and > maybe others) and since they are used on our environment, now some > applications doesn't work any more. > > Is there a way to get these variable set again ? > > Thank you > Marco Induni Hi Marco, I have a related bug about this, but I would really like to know exactly what are the needed variables that are now missing. Could you run some test and provide me with this info? That would help a lot. Thanks (In reply to Felip Moll from comment #2) > (In reply to Marco Induni from comment #0) > > Dear support, > > I followed the instructions (https://slurm.schedmd.com/pam_slurm_adopt.html) > > to limit the access on the compute nodes where there is a JOB running and it > > works. > > I have only a problem discovered, because some environment variables now > > aren't set anymore. > > Indeed based on the instruction you should comment out in the > > /etc/pam.d/system-auth the following > > > > #account sufficient pam_localuser.so > > #-session optional pam_systemd.so > > > > The problem is that pam_systemd set for example all the XDG variables (and > > maybe others) and since they are used on our environment, now some > > applications doesn't work any more. > > > > Is there a way to get these variable set again ? > > > > Thank you > > Marco Induni > > Hi Marco, > > I have a related bug about this, but I would really like to know > exactly what are the needed variables that are now missing. > > Could you run some test and provide me with this info? > That would help a lot. > > Thanks Dear Felip, as a test I logged on a node with the configuration adopt enabled and a second time with the adopt disabled and these are the missing variables: XDG_RUNTIME_DIR=/run/user/22447 XDG_SESSION_ID=8863 If you need that I have to run a specific test or you need something else, just let me know. Thank you Marco Any update on this ? Kind regards, Marco I. (In reply to Marco Induni from comment #4) > Any update on this ? > > Kind regards, > Marco I. Hi Marco, Not yet. The issue is not easy. The pam systemd module just modifies the cgroup and overwrites what the pam slurm does. If you remove pam systemd these variables are not set, and we cannot recreate them because the values of the session ID comes from a new session generated by the pam systemd module. I am studying again what options do we have. Will inform you as soon as we've any conclusion. Hi Marco, I am marking this bug as a duplicate of bug 5920. We are discussing there a fix for this issue which would solve your situation. Please, add yourself to bug 5920 and feel free to reply there if you have questions. *** This ticket has been marked as a duplicate of ticket 5920 *** |