v.1.2.1

• basic topologyplugin support
• GRES
• travis automatic testing is now working (should catch most template errors)
• maxarraysize can be set
• set some sysctl settings according to the HTC guide for slurm

v1.2.0

only make new NIS DB if add user or group changed
Use MAX_TRES by default
Allow setting ControlAddr
Pack serial jobs at end by default
Allow setting ControlAddr
add the variable introduced in #31 to defaults (but commented out)
mention separate db host in readme

v1.1.0

Four PRs from @jabl:

• #25 - don't force to have the slurmdb host on the same as the slurmctld (but it defaults to this)
• #26 - only use the default partitions if the slurm_partitionlist variable is set
• #27 - cgroup and a variable name typo
• #28 - allow setting priority flags for the multifactor

v1.0.0

Many changes.

• use cgroups
• use ansible lists to create nodes and partitions
• uid / gid consistency for munge and slurm user
• possibility to have per job /tmp directories
• restrict which users can ssh into a node
• slurmdbd logs to syslog

v0.9.0

There's been a big cleanup

• lots of settings in slurm.conf are now ansible variables
• Install rpms from FGCI yum repo
• log to syslog
• slurm health check - defaults to NHC
• where to download munge.key from

v0.1.0

First public release.
This builds slurm and distributes it with ansible/scp.
In a release in the near future we'll take this out / at minimum not make it the default.