Sbatch: error: Batch job submission failed: Unable to contact slurm controller (connect failure)

HPC Technical
1

Hello,

I cannot submit jobs using sbatch on Yggdrasil:
sbatch: error: Batch job submission failed: Unable to contact slurm controller (connect failure)

Salloc seems to work.

Could you please help me?

Thank you!

2

I am still having problems with the slurm controller… any news?

3

For future reference, I found the cause of the error: bad job configuration file.

I copied the config from baobab, but apparently some of the arguments were incompatible.
I suspect I was asking too many cores on a gpu node, but I did not investigate further.

I am still a bit surprised by the SLURM error itself (“connect failure”) tho…

4

Hi,

Maybe you specified the Baobab cluster in your sbatch and you submited your job from Yggdrasil? This isn’t possible (yet!)

5

Hey Yann,

I think you are right. I did not specify baobab, but I did use “yggdrasil”!

I thought the problem was the number of cores, but that raises a different error.

I now understand why it complains about the connection… I assume the cluster “yggdrasil” does not exist, right?

6

The cluster yggdrasil do exist but the configuration we are using is incorrect when you try to use the pragma #SBATCH --cluster=XX. Please do not specify the cluster name in your sbatch/srun.

Best

Yann

7

Sure, I corrected all my scrips. Thanks!

8

Hi, I am facing the same error.

sbatch: error: Batch job submission failed: Unable to contact slurm controller (connect failure)

My job does not contain the specifications mentioned above:

#!/bin/sh
#SBATCH --partition=public-cpu
#SBATCH --time=00:30
#SBATCH --mail-user=adriano.rutz@unige.ch
#SBATCH --mail-type=ALL

ml GCC/9.3.0 Singularity/3.7.3-Go-1.14

#convert job array index to three digit padded with zeros
printf -v FILE_INDEX "%04d" ${SLURM_ARRAY_TASK_ID}

FILE=test/test-${FILE_INDEX}.txt

srun singularity run cfm-4/cfm.simg -c "cfm-predict $FILE 0.001 /trained_models_cfmid4.0/[M+H]+/param_output.log /trained_models_cfmid4.0/[M+H]+/param_config.txt 1"
9

Hi Adriano,

We have updated slurm configuration on baobab at the same time you submitted.

Could you try again and avise me if it’s still not working ?

10

Hi Adrien,

sbatch --array=1-10 run_cfm_test.sh
Submitted batch job 56373639

Thank you!