Are you utilizing message aggregation?

On 19/09//2016 17:52, Ulf Markwardt wrote:
Dear all,

we have about 2000 nodes running with Slurm 16.05.4 (per-core
granularity). Under heavy load, the slurmctld is
sending messages to the nodes with a transfer rate between 50 and 80
MByte/s (double-checked the ports with tcpdump and a small script).

a) Is this normal?
b) What information is this daemon transmitting?
c) It seems like the slurmstepd are listening to this, correct?
d) How can we reduce the amount of communication.


Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to