Stats page fires 63 ceilometer-api requests, runs into timeout

[Peter Sabaini]

When accessing "Resources Usage Overview" / "Stats" I get timeouts for Period=weekly.

I've now traced the line between web frontend and ceilometer API when accessing a stats page ("memory", "last week"). There are about 63 HTTP requests going through. When running those via curl, this takes just a little under 4min. The data is coming back fine, but I suspect the web UI is not patient enough for that. Total data transferred is ~1.4M (uncompressed).

[James Page (james-page)]

I'm not entirely sure what we can do about this; performance of the calls is likely to be driven by the speed of the underlying hardware supporting ceilometer and mongodb. This does not sound like a charm specific problem, so I'm going to push this to the Ubuntu package - it would be useful if you can provide details as to which version of OpenStack and Ubuntu where being used when you saw this issue.

[James Page (james-page)]

Once we have a little more information (and maybe confirmation that this is still a problem on a recent, still supported upstream version of openstack) we can see if this appropriate to push to the horizon project directly.

[Peter Sabaini (peter-sabaini)]

I've checked this behaviour with a Mitaka install, and this specific issue (many small http requests timing out) is not present anymore. From my POV this bug can be closed.

[Launchpad Janitor (janitor)]

[Expired for horizon (Ubuntu) because there has been no activity for 60 days.]