Feature request: prometheus metrics to track Apache 503s for OpenStack services
Bug #2020561 reported by
Paul Goins
This bug affects 1 person
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| OpenStack Keystone Charm |
Triaged
|
Wishlist
|
Unassigned | ||
Bug Description
I'm filing this bug in response to a case on one of our internal clouds.
We had a case where the WSGI workers for the keystone service were apparently getting intermittently overloaded. This was observed in the form of "503 Service unavailable" errors.
This was mitigated by changing the worker-multiplier on the keystone app from 0.25 to 0.375.
It was raised as a concern that we may want a Prometheus metric for tracking these 503 errors. Can this be considered?
Revision history for this message
| Paul Goins (vultaire) wrote | #1 |
| Changed in charm-keystone: | |
| status: | New → Triaged |
| importance: | Undecided → Wishlist |
To post a comment you must log in.
Obviously this may apply to many of the OpenStack charms, but I filed it against the keystone charm since that's where we concretely observed the issue.