1
LLNL-SPECIFIC RELEASE NOTES FOR SLURM VERSION 2.0
4
For processor-scheduled clusters (*not* allocating whole nodes to jobs):
5
Set "DefMemPerCPU" and "MaxMemPerCPU" as appropriate to restrict memory
6
available to a job. Also set "JobAcctGatherType=jobacct_gather/linux"
7
for enforcement (periodic sampling of memory use by the job). You can change
8
said sampling rate from the default (every 30 seconds) by setting the
9
"JobAcctGatherFrequency" option to a different number of seconds in
12
For InfiniBand switch systems, set TopologyType=topology/tree in slurm.conf
13
and add switch topology information to a new file called topology.conf.
14
Options used are SwitchName, Switches, and Nodes. The SwitchName is any
15
convenient name for bookkeeping purposes only. For example:
16
# Switch Topology Information
17
SwitchName=s0 Nodes=tux[0-11]
18
SwitchName=s1 Nodes=tux[12-23]
19
SwitchName=s2 Nodes=tux[24-35]
20
SwitchName=s3 Switches=s[0-2]
22
Remove the "preserve-env.so" SPANK plugin. The functionality is now
25
SLURM version 2.0 must use a database daemon (slurmdbd) at version 2.0
26
or higher. While we are testing version 2.0, set "AccountingStoragePort=????".
27
Once we upgrade the production slurmdbd to version 2.0, this change will
28
not be required. You can likewise test 1.3.7+ clusters with the same port
29
since 2.0 slurmdbd will talk to 1.3.7+ SLURM.
31
SLURM state files in version 2.0 are different from those of version 1.3.
32
After installing SLURM version 2.0, plan to restart without preserving
33
jobs or other state information. While SLURM version 1.3 is still running,
34
cancel all pending and running jobs (e.g.
35
"scancel --state=pending; scancel --state=running"). Then stop and restart
36
daemons with the "-c" option or use "/etc/init.d/slurm startclean".