Detect thrashing using PSI

[hakavlad]

IMHO PSI is maybe best metrics to detect thrashing.
https://lwn.net/Articles/759658/
https://facebookmicrosites.github.io/psi/

You can try to use it to detect thrashing instead of vmstat.

PSI file example (/proc/pressure/memory):

some avg10=70.24 avg60=68.52 avg300=69.91 total=3559632828
full avg10=57.59 avg60=58.06 avg300=60.38 total=3300487258

Use total metrics.

[tobixen]

This is interesting.

This sounds a bit like thrash-protect to me:

We wrote a <100 line POC python script to monitor memory
pressure and kill stuff way before such pathological thrashing leads
to full system losses that require forcible hard resets.

At some point I should look more into this and replace the current algorithm (based in swapin*swapout) with the new /proc/pressure/memory-statistics. However, /proc/pressure is missing in most of the production servers I'm having responsibility for, so it will take a long time before the backward-compatible algorithm can be obsoleted.