[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[HTCondor-users] increasing schedd memory usage [v8.6.0?]



Hi all,

we are observing since yesterday a suddenly increasing memory usage on
one of our condor schedulers [1].
Several condor schedds processes start allocating each ~35% of memory
(16gB overall). Judging from the (coarse grained (1m)) monitoring,
memory usage is oscillating with ~5m between 50% and >100% and swapping.
After restarting the service/node, memory consumption got down
significantly but peaked again later during the day.

So far we have not found a cause for the sudden hunger for memory.

On Feb. 01 we upgraded the node from 8.4.1 to 8.6.1 and kept its sibling
node on 8.4.1. So far, the 8.4.1 sibling has not shown the behaviour. On
the other hand it took the 8.6.1 node more than week to get into the
current state, so it is imho not very conclusive to suggest the update
as cause? [2]

Maybe somebody has an idea, where to look for clues?

Cheers and thanks for ideas,
  Thomas


[1]
grid-arcce1.desy.de - Nordugrid ARC CE submitting to Condor

> free
             total       used       free     shared    buffers     cached
Mem:      16268636    8605212    7663424        308     423620     551716
-/+ buffers/cache:    7629876    8638760
Swap:      4095996    1308184    2787812

[2]
in the release notes I found nothing matching that could point to our
observations
http://research.cs.wisc.edu/htcondor/manual/v8.6/10_3Stable_Release.html

Attachment: grid-arcce1_mem_3d_201702091735.png
Description: PNG image

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