High CPU & Memory usage when several providers are configured

[Subarash11]

We have more than 20 OpenShift + 3 OpenStack providers configured on a Region managed by 2 Petrosian servers : 1 UI server + 1 Worker.
The Worker is using more than 40 GB Memory and the 6 cores CPU usage is regularly high.
Theses OpenShift and OpenStack providers are only used for reports.

Did you know an issue to reduce Memory and CPU usage?
Is it maybe possible to limit monitoring updates to once a day?

[Fryguy]

@agrare Thoughts here? I think the real way forward is to start moving to non-Rails workers but I'm curious if there is a something that can be done in the interim.

[agrare]

Yeah moving away from rails workers should help with the memory, not with the CPU though. 20 openshift clusters is a pretty decent size deployment for 1 appliance.

Is it maybe possible to limit monitoring updates to once a day?

This is interesting, if you don't care about "real-time" up to date inventory it could be possible to disable the RefreshWorker and only run scheduled refreshes.

[Subarash11]

@agrare
I tried to modify configuration file to Settings > ManageIQ Region > Advanced
Then, I changed "rails_worker" to false and I changed poll to 24.hours instead of 10.seconds behind "ems_refresh_worker" option.
After restarting evmserverd it doesn't reduce Memory usage.
Is is the correct way I followed?

[Fryguy]

rails_worker is only implemented for the vmware provider, so it won't make a difference. Setting poll to 24 hours should reduce the CPU usage, but probably not a whole lot.