#1526 [HHQ-2072] One minute collection interval only updating data

[Hyperic Team]

http://jira.hyperic.com/browse/HHQ-2072
I've run into an issue with a custom metric that pops up depending on when the agent polls the data. When the HQ Agent grabs data for my metric (and probably all others that run on a minute granularity) and the time it takes to collect the metrics crosses over a minute boundary on the system clock, it won't collect new data the next minute. This probably sounds confusing so I'll attach an example of one that's working perfectly fine and another that isn't for the same metric on different servers.

I modified my metric script so that it would echo when it starts and ends.

SERVER THAT IS WORKING (screenshot WORKING):


START: Wed Apr 23 21:22:23 CDT 2008


END: Wed Apr 23 21:22:25 CDT 2008


START: Wed Apr 23 21:23:23 CDT 2008


END: Wed Apr 23 21:23:25 CDT 2008


START: Wed Apr 23 21:24:23 CDT 2008


END: Wed Apr 23 21:24:26 CDT 2008


START: Wed Apr 23 21:25:23 CDT 2008


END: Wed Apr 23 21:25:25 CDT 2008


START: Wed Apr 23 21:26:23 CDT 2008


END: Wed Apr 23 21:26:25 CDT 2008


START: Wed Apr 23 21:27:23 CDT 2008


END: Wed Apr 23 21:27:25 CDT 2008


START: Wed Apr 23 21:28:23 CDT 2008


END: Wed Apr 23 21:28:25 CDT 2008


START: Wed Apr 23 21:29:23 CDT 2008


END: Wed Apr 23 21:29:26 CDT 2008

You'll notice that this server is working properly and collecting every minute. It's also starting on the 23rd second of every minute. The graph that I have attached also shows fluctuated data every minute.

SERVER THAT IS NOT WORKING (screenshot NOT_WORKING):


START: Wed Apr 23 21:23:59 CDT 2008


END: Wed Apr 23 21:24:01 CDT 2008


START: Wed Apr 23 21:25:59 CDT 2008


END: Wed Apr 23 21:26:01 CDT 2008


START: Wed Apr 23 21:27:59 CDT 2008


END: Wed Apr 23 21:28:01 CDT 2008


START: Wed Apr 23 21:29:59 CDT 2008


END: Wed Apr 23 21:30:01 CDT 2008


START: Wed Apr 23 21:31:59 CDT 2008


END: Wed Apr 23 21:32:01 CDT 2008

You'll notice here that it's only running the script every 2 minutes. I'm assuming that this is because it's crossing the minute boundary. If you look at the graph for this one, it's reporting the same exact number for each minute in the two minute period.

I have successfully been able to resolve the issue on a server by stopping and starting the agent, thus changing the start time that the agent polls the data, but I'd rather have a permanent fix if one is available.

See attached support ticket for ScreenShots