We recently added a test for the ovs-vswitchd log messages, however it turns out they're extremely common up to even 9s, especially on gcp. The test only ever flakes.

It would be good to plot these on the intervals chart, however there's no category they fit into yet. This is a sysmted journal message. This should only be done on the sippy side where we're going, no need to do it multiple times.

Also we should extract the test metadata in sippy, so we can query the actual time values involved. See testoutputmetadata.go for examples on how to do this. Then we could chart how it was behaving.

Extract and chart the user/system time from the parenthesis as well.

Sample log:

node/ci-op-2dm3ldyt-0dd98-rbs8n-worker-centralus3-w48ph - ovs-vswitchd[1219]: ovs|00001|timeval(revalidator14)|WARN|Unreasonably long 1530ms poll interval (0ms user, 1ms system)

So we would want metadata for "intervalMS", "userMS" and "systemMS".