Hi Ricardo,
I am making some progress into the installation and setup. For
compatibility reasons, I have a need to run the slurm on CentOS7. Few
comments on my side:
minor adjustment is needed for the my.cnf file in the
elasticluster/share/playbooks/roles/slurm-master/tasks/db.yml
- name: Ensure InnoDB parameters are large enough for SLURM DBD (Ubuntu)
tags:
- slurm
- slurmdbd
blockinfile:
# FIXME: Path is correct on Debian/Ubuntu; is it also for CentOS/RHEL?
path: '/etc/mysql/my.cnf'
state: present
backup: yes
insertafter: EOF
content: |
# See https://wiki.fysik.dtu.dk/niflheim/Slurm_database#id5
[mysqld]
innodb_buffer_pool_size=1024M
innodb_log_file_size=64M
innodb_lock_wait_timeout=900
when: 'is_debian_or_ubuntu and not (is_debian_8_or_later or
is_ubuntu_14_04_or_later)'
- name: Ensure InnoDB parameters are large enough for SLURM DBD (RHEL)
tags:
- slurm
- slurmdbd
blockinfile:
# FIXME: Path is correct on Debian/Ubuntu; is it also for CentOS/RHEL?
path: '/etc/my.cnf'
state: present
backup: yes
insertafter: EOF
content: |
# See https://wiki.fysik.dtu.dk/niflheim/Slurm_database#id5
[mysqld]
innodb_buffer_pool_size=1024M
innodb_log_file_size=64M
innodb_lock_wait_timeout=900
when: 'is_rhel7_compatible'
(alternatively, the /etc/my.cnf can be softlinked to /etc/mysql/my.cnf)
Additionally, default playbook does not set the permissions of the
/etc/slurm/slurmdbd.conf to 600. Upon brief examination of the logs :
slurmdbd[15118]: fatal: slurmdbd.conf file /etc/slurm/slurmdbd.conf should
be 600 is 444 accessible for group or others
(did not have the chance to find which playbook would be responsible for
that, after manual fix, re-run of the elasticluster.sh was successfull.
There are minor errors related to FastSchedule and Openmpi:
Feb 15 22:53:17 slurm5-frontend001 slurmctld[15051]: error: Ignoring
obsolete FastSchedule=1 option. Please remove from your configuration.
Feb 15 22:53:17 slurm5-frontend001 slurmctld[15051]: error: Translating
obsolete 'MpiDefault=openmpi' option to 'MpiDefault=none'. Please update
your configuration.
Apart from that, cluster is up and ready for testing.
I am going to be exploring slurm on openstack over the next couple of
weeks, if need to test anything, happy to help.
Kind Regards,
Tomasz
--
You received this message because you are subscribed to the Google Groups
"elasticluster" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticluster/903a5def-db96-4121-875f-3e3968a7e447n%40googlegroups.com.
