Sbatch: error: s_p_parse_file: unable to read "/etc/slurm/slurm.conf": Permission denied

Get Help
1

Hello,

I got error when I submit an interactive desktop.

image
image1194×219 67.9 KB

my slurm.conf is readable for everyone.

# ls -l /etc/slurm/slurm.conf
-rw-r--r--. 1 root root 7922 Jul 28 14:06 /etc/slurm/slurm.conf

Indise of slurm.conf, ClusterName is set.

ondemand-web, slurm controller and cluster host are different but the communication between them are okay.

/etc/ood/config/clusters.d/my_cluster.yml is below.

---
v2:
  metadata:
    title: "My Cluster"
  login:
    host: "hpc-lg01.my_cluster.com"
  job:
    adapter: "slurm"
    bin: "/usr/bin"
    conf: "/etc/slurm/slurm.conf"
  batch_connect:
    basic:
      script_wrapper: |
        module purge
        %s
      set_host: "host=$(hostname -A | awk '{print $1}')"
    vnc:
      script_wrapper: |
        module purge
        export PATH="/opt/TurboVNC/bin:$PATH"
        export WEBSOCKIFY_CMD="/usr/bin/websockify"
        %s
      set_host: "host=$(hostname -A | awk '{print $1}')"

and /etc/ood/config/apps/bc_desktop/my_cluster.yml

---
title: "Cluster Desktop"
cluster: "my_cluster"
attributes:
  desktop: "xfce"

I might miss something but I couldn’t find it.

I already test interactive desktop with dev-cluster and it worked well. But, with my_cluster, it does not work.
Thank you in advance for any help.

PARK

2

I see this as 644 - but I wonder about the directory above it. Can you ls -l as a regular unprivileged user? Unable to read/permission denied is pretty straight forward.

Remember to ls -l on the OnDemand machine itself, this is where the command is being issued, so this is where you have to test to validate it.

3

Thank you for your answer.
Here it is for the permission.

s.park@ondemand-web01 ~ $ ls -l /etc/slurm/
total 16
-rw-r--r--. 1 root root 7922 Jul 28 14:06 slurm.conf

PARK

4

Maybe it’s an SELinux thing?

Can you just issue sbatch with and without CONF environment variable set to that /etc file and see if it works from a command line?

5

With slurm.conf,

This is script file for sbatch

#!/bin/bash
echo "Hello world!"

I submitted this job with sbatch.

s.park@ondemand-web01 ~ $ sbatch tx.sh
Submitted batch job 948700

and output is..

s.park@ondemand-web01 ~ $ cat slurm-948700.out
Hello world!

Without slurm.conf
I got failed.

s.park@ondemand-web01 ~ $ sbatch tx.sh
sbatch: error: resolve_ctls_from_dns_srv: res_nsearch error: Unknown host
sbatch: error: fetch_config: DNS SRV lookup failed
sbatch: error: _establish_config_source: failed to fetch config
sbatch: fatal: Could not establish a configuration source

Also, as expected, without slurm.conf, tired to submit desktop on the web and failed with this error.

image
image1197×229 72.5 KB

PARK

6

Not without the file itself, with and without the CONF=/etc/slurm/slurm.conf environment variable set.

7

I’ve comment the conf line in cluster .yml file as below.

 job:
    adapter: "slurm"
    bin: "/usr/bin"
      #    conf: "/etc/slurm/slurm.conf"

However, with or without conf environment variable, same error is occured.

PARK

8

It’s likely an SELinux issue. Do you have SELinux enabled?

1 Like
9

And if so, check out /var/log/audit.log and see if there are errors there.

10

Yes, SELinux was enforced.
I disabled it and then it works well.
Thank you for your guide.

PARK

11

Pull requests to this component are welcome! It’s always slightly different on different systems, so any patches we get on it from the community are greatly appreciated!

1 Like