Hi folks,
We've just upgraded to slurm 17.02.9 (native) on our Crays, but can't
get sinfo to work on them anymore from a non-cray
"sinfo: error: Cluster 'galaxy' has an unknown select plugin_id 108"
On the Crays we have
aelwell@galaxy-int:~/testjobs/native$ grep -i select /etc/opt/slurm/slurm.conf
SelectType=select/cray
SelectTypeParameters=CR_ONE_TASK_PER_CORE,CR_CORE_Memory,other_cons_res
aelwell@galaxy-int:~/testjobs/native$
and on the non-cray node I'm trying to get working (an admin node we
use for monitoring the job Qs across the site)
hpc-admin2:~ # zypper up
Refreshing service 'SMT-https_target_pawsey_org_au'.
Loading repository data...
Reading installed packages...
Nothing to do.
hpc-admin2:~ # rpm -qa | grep slurm
slurm-munge-17.02.9-6.10.1.x86_64
slurm-17.02.9-6.10.1.x86_64
slurm-plugins-17.02.9-6.10.1.x86_64
hpc-admin2:~ # ldd /usr/lib64/slurm/select_cray.so
linux-vdso.so.1 (0x00007ffdf9db1000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007fcd37d89000)
libc.so.6 => /lib64/libc.so.6 (0x00007fcd379e6000)
/lib64/ld-linux-x86-64.so.2 (0x00007fcd381b9000)
hpc-admin2:~ # sinfo --version
slurm 17.02.9
hpc-admin2:~ # sacctmgr show clusters | grep gala
galaxy 146.118.55.132 6817 7936 1
normal
hpc-admin2:~ # sinfo -M galaxy
sinfo: error: Cluster 'galaxy' has an unknown select plugin_id 108
sinfo: error: 'galaxy' can't be reached now, or it is an invalid entry
for --cluster. Use 'sacctmgr list clusters' to see available
clusters.
hpc-admin2:~ #
Is there a workaround to get non-crays talking to cray slurmctlds?
Many thanks,
Andrew
