-
30. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 9:59 AM (in response to amit8484)Jboss Application Server DOWN
ID : 102896
Parent ID : 102852
Resource ID : 512809
Recover ID : 0
Jboss Application Server Recovery Alert
ID : 103052
Parent ID : 102998
Resource ID : 512809
Recover ID : 102852 -
31. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 10:00 AM (in response to amit8484)I'll believe you if you say it isn't working, but I still need to know whether it's not working because the definitions are wrong or because of some other internal issue. My last instructions should help clarify that, and help me to track down the root cause.
-
32. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 10:02 AM (in response to amit8484)i posted the results above, you can see that the recovery id is still looking for the parent id, which would be the template... right?
-
33. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 11:00 AM (in response to amit8484)Correct, it looks like the recovery id is still pointing to the template and not the definition? Did you change the template AFTER running the script? If so, you need to re-run the script. Can you confirm that the output you're showing me is what you get immediately after running the SQL?
-
34. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 11:07 AM (in response to amit8484)ill have the person re-run the script right now than ill test again
-
35. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 1:22 PM (in response to amit8484)well i re ran the query, and i tested it again, but now the original (jboss apps server) alert doesnt fire at all!
-
36. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 1:23 PM (in response to amit8484)Is it still disabled from the last firing?
-
37. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 1:25 PM (in response to amit8484)After re-running the query, do the recovery ids on the resource-level look correct now?
-
38. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 1:28 PM (in response to amit8484)lemme check
-
39. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 2:24 PM (in response to amit8484)okay, it seems to match correctly now, but now the original alert is not firing, let me test again, ill get back to you
-
40. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 2:32 PM (in response to amit8484)You very likely need to re-enable the first alert definition manually. Since the recovery alert wasn't set up properly at the time the first alert fired, it was disabled and there was nothing to re-enable it.
-
41. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 2:52 PM (in response to amit8484)this is really wierd. Okay an alert fired, but when i ran the match check again by running the query, it reverted back to matching with the parent id! so before the alert fired, everything was matching correctly, but after the alert fired, it went back to matching with the parent ID!
-
42. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 3:02 PM (in response to amit8484)okay, nm, thats didnt happen, but the recover alerts still arent working, everything is matched up though
-
43. Re: could this be whwy my recovery alerts arent working?
joe.marques Jun 22, 2009 3:19 PM (in response to amit8484)OK, one thing that might be at play here is the fact that direct SQL statements against the alerts-related tables do not immediately update the in-memory alert condition cache. Go to the services tab of the resource browser and filter the results for resources of type "RHQ Server Alerts Engine Subsystem"
You'll have one type of this resource for each Jopr server you have in your enterprise. Go to the operations tab and execute the "Reload Caches" operation. This will sync the in-memory alerts condition cache with the correct data in the database.
To recap:
* each time you edit any alert template, re-execute that SQL in RHQ-2150
* find the "RHQ Server Alerts Engine Subsystem" resource and execute "Reload Caches" operation
Things should work after that.
Note: I'm raising the priority of this bug in our issue tracker - I'm trying to a workaround here, but it's proving to be too tedious and repetitive, and thus error-prone. As soon as I commit the fix to Jopr svn, I'll let you know and you can check out the code from source to see the results. -
44. Re: could this be whwy my recovery alerts arent working?
amit8484 Jun 22, 2009 3:32 PM (in response to amit8484)hmmm.. i see the RHQ server alerts engine subsystem but its not available and no operations are available