Red Hat Bugzilla – Bug 839583
utilisation graph is slow
Last modified: 2013-01-16 23:34:37 EST
Description of problem:
utilisation graph has default date range 2009-05-21 - today.. maybe that is the reason why it takes so much time to load. I think most of the people are interested in shorter period, could it be e.g. last month/week by default?
Version - 0.9.0
The code renders the same number of points in the graph no matter what date range is selected. Its slowness is correlated with number of systems in the data set, not date range. Unfortunately there is just no way to efficiently produce the data needed from Beaker's db schema.
The best approach is probably to abandon it altogether and use Graphite or some other time series data store instead:
I think we should include some *documented* recommendations for the metrics stuff we're farming out to Jasper and Graphite.
For this one, since the code to send notifications to Graphite is already in place, we can just recheck it this release cycle.
However, we should also figure out some specific suggestions for Graphite links to add via the external reference feature being added in #883606 and include those in the documentation. This can be part of the documentation of the Graphite support in the Admin Guide, which I don't believe exists yet...
Docs for Graphite integration:
System utilisation metrics, to match up with the existing SQL-based utilisation graph (the original complaint in this bug):
Suggested URL in the docs was wrong:
Can't really verify docs until they are on a website somewhere, and cannot verify graphite metrics until they are being sent.
Built website (incl docs) locally, verified metrics on personal graphite server.
Beaker 0.11.0 has been released.