Hide Forgot
Description of problem: In running scale tests we had witnessed an instance where the host info zeros out. Version-Release number of selected component (if applicable): Development How reproducible: ? Steps to Reproduce: Unknown Actual results: See attachment Expected results: Normal Data
Ok, I've done some poking around on this one. Everything that I've seen so far seems to point to this one being a symptom of a condor_restart. After a condor_restart, the first update from the Collector gives us zeroes for all the host info. In my testing, I always got the correct data by the second update, but I can imagine a setup where you may be subject to a few update periods of either zeroes or values less than the full host totals. I'm inclined to chalk this one up to expected behavior, but if anyone has anything to add, I'm all ears.
COLLECTOR_UPDATE_INTERVAL defaults to 15 minutes, set it to a lower value (say 30 seconds) to get smoother data.