( btw. Jopr is phased out - RHQ is the new kid in town -- see http://rhq-project.org/ , which contains all the Jopr bits. )
How do you define RealTime? If you need second or sub-second measurement data e.g. like in driving a production plant, then it is definitively the wrong tool.
As you saw 30sec intervals are the smallest we allow. With larger installations with thousands of resources and ten thousands of metrics, this still produces huge amounts of data.
In the bars, the diamond shows the average. That occurs if the bar represents more than one measurement taken. So the bar itself shows the lowest and highest values during that time and the diamond the average.
You may want to modify the display timeframe at the bottom of the page to select a shorter timeframe to view, which allows you to drill down into the individual values.
thanks for your prompt reply, !!
Yes the system im currently working in, demands monitoring of data with max 5 or 10 second time frame. ( Since jopr only allows 30secs - I suppose its not the right tool then :'(
Also Does jopr allow Pie charts or any other charts except for bar charts ?
Also i read Nagios is coming soon to RHQ. Does that mean with Nagios it Jopr / RHQ will be able to give me second or sub second measurement ?
Thanks as always
while the UI is limiting the setting to 30sec, you could try via modification of db tables (or via CLI ?) to set the metric taking to 10s and see how RHQ reacts. If you only have a limited number of metrics to collect that fast, if could even work, but this is like driving without seat belt :-)
In RHQ 3 we only have the kind of graphs we have. In RHQ 4 we may/will see different ones (you want to contribute? :-)
Nagios coming to RHQ means that you monitor your stuff in Nagios and RHQ can depending on RHQs schedule take the data from Nagios and then e.g. alert on it. THe 30sec rule still applies for the data intake from Nagios.
Your tip seems really great..!! I'll try it and update you on it.
to the other question , yes i would like to contribute to the community, in which way i could do that.
I really dont want to switch to any other tool for monitoring .
which particular table would i need to edit to change the metric to 10 sec or whatever interval i want ??
check the following page about contributing to RHQ: http://www.rhq-project.org/display/RHQ/Contributions
The table is RHQ_Measurement_sched
I've just checked with RHQ master branch and there you can have live data views of graphs when you click on a little icon next to the 'usual' graph. Perhaps this is something for you to follow?
Note that we are switching the UI from struts/jsp and jsf to GWT, so any patches to struts or jsf pages may get outdated the moment you send them :-)
In the rhq_measurement_sched table i see different columns, the one that has to be edited is which one ?
Also, what does mtime and coll_interval signify ? I see under the hq_measurement_schd table, column mtime doesnt seem to have any value .
Also with my guess coll_interval was the column which i edited in rhq_measurement_sched. i changed the value to 10000 (10sec) , and when i started the rhq server again, even after setting the time interval to 10secs the values dont change in "collection Interval column" in Schedules in jopr server
Please help me with this,
Just a reminder that RHQ should not be considered a realtime monitoring solution. Yes you probably can collect a FEW metrics every 10 seconds, but collecting ALOT of metrics every 10 seconds would probably stress the system.
Also, the agent does NOT send metrics up to the server as soon as it collects them. The agent will check to see if it has any metrics it needs to send to the server every 30 seconds - this has nothing to do with the individual metric collection intervals. So even though you might be able to have the agent collect individual metrics every 10 seconds, the server won't actually get them that fast. At most it will get them every 30 seconds (because, as I said, the agent sends up measurement reports every 30 seconds at most). Here's the code that shows this - this is from the agent's MeasurementManager:
// Schedule the measurement sender to send measurement reports periodically.
this.senderThreadPool.scheduleAtFixedRate(measurementSenderRunner, collectionInitialDelaySecs, 30,
Unlike most other things in the agent, this is not configuable, its hardcoded at 30 seconds - not something I personally like but that's the way it is.
As for the table columns, "mtime" is the "modified time" - that is, the last time that schedule was modified (for example, the last time the collection interval was changed) - this is in epoch milliseconds. coll_interval is the collection interval (in milliseconds I believe) - this would be set to your 10 second time (if the units is in milliseconds, which I think it is, then it would be 10000). Just changing this column might not affect the agent though - we have agent synch'ing going on under the covers - the agent synchronizes its internal data with the server when the agent starts up. It would have to know to sync its collection intervals with the server - you'd have to change some other information about the agent via the GUI to ensure the agent will sync properly. I do not know why the UI itself doesn't show the 10s - perhaps you changed the wrong row in the table?
This is getting well beyond what we recommend doing with the software - if you want to go further, I recommend you walk through code and learn what its doing under the covers to figure out if you can do what you want.
For your detailed reply,
I want to give this one last shot. Do you which jar file would contain the hardcoded value "30 seconds " ?
Where is MeasurementManager located ? I cant find the in rhq-agent/lib/ directory ?
This is a core PC (plugin container) class - hence, its in the rhq-core-plugin-container jar.