These are chat archives for ManageIQ/manageiq/performance

24th
Sep 2018
Marek Aufart
@aufi
Sep 24 2018 11:39
@Ladas Hi, I talked with @psuriset , I experienced very incosistent OpenStack response times (with gnocchi metric/resource list) I'm not sure if it can be an issue, but it looks strange to me. I checked all 3 controllers and cannot see an overloaded node, would you have an idea for a health check on OpenStack side (I haven't found anything obvious, hopefully someone from OpenStack teams migth check it)
Ladislav Smola
@Ladas
Sep 24 2018 11:40
@aufi so, not sure about the metrics, I've seen some timeouts
@aufi for the refresh, there seems to be a but with heathbeat
@aufi there should be 10s + 120.minutes timeout, but somehow it's only 130s
Marek Aufart
@aufi
Sep 24 2018 12:02
Metrics were timeout from OSP side (default 60s), which, I think, is possible with 100s of instances, I suggested increase that timeout to 120s
about refresh, so it seems that it gets killed on MIQ side sooner, than it finishes?
Ladislav Smola
@Ladas
Sep 24 2018 12:03
@aufi is the metrics timeout configurable on our side?
@aufi yeah, I am trying to debug the refresh now
Marek Aufart
@aufi
Sep 24 2018 12:13
metrics timeout is part of read_timeout option
for all OpenStack calls
Daniel Berger
@djberg96
Sep 24 2018 18:23
@Ladas Just discussing batch saves for AR got me looking around, came across this: https://hmistry.github.io/software/2017/12/11/activerecord-vs-sequel.html