[#RHQ-849] server-specific jobs only seem to be running on one server in the cloud, instead of on EACH server

RHQ Project

server-specific jobs only seem to be running on one server in the cloud, instead of on EACH server

Created: 17/Sep/08 11:52 AM Updated: 29/Sep/08 03:02 PM

Component/s:

Core Server, High Availability

Affects Version/s:

1.1pre

Fix Version/s:

1.1

Time Tracking:

Not Specified

Resolution Date:	17/Sep/08 04:29 PM
Date of First Response:	29/Sep/08 03:02 PM
Tester:	Jeff Weiss
VCS Revision:	1,527

Description

« Hide

this was seen by greg. you can verify this by starting up a 2-server cloud and going to the List HA Servers page. if the bug is present, one of them will be marked as DOWN incorrectly.

All

Comments

Work Log

Change History

Sort Order:

Ascending order - Click to sort in descending order

[ Permlink | « Hide ]

Joseph Marques - 17/Sep/08 11:53 AM

rev1516 - making the job group unique seems to make each server execute their own instance of it;

[ Permlink | « Hide ]

Joseph Marques - 17/Sep/08 01:33 PM

rev1519 - make sure server status is reset to NORMAL when it beats to the database with a new mtime;
add some debug logging to the ClusterManagerJob and ServerManagerJob;

[ Permlink | « Hide ]

Joseph Marques - 17/Sep/08 01:47 PM

rev1520 - make server-specific jobs volatile...this will hopefully prevent other servers in the quartz cluster from picking up the responsibility of executing those jobs in failover conditions;

[ Permlink | « Hide ]

Joseph Marques - 17/Sep/08 04:22 PM

rev1526 - use the EJB3 timer service whenever we want to perform periodic, server-specific tasks (under the covers it's using the RamJobStore for a non-clustered quartz instance, precisely the desired semantics);
update the CacheConsistencyManager and ServerManager to leverage EJB3 timer service;

[ Permlink | « Hide ]

Joseph Marques - 17/Sep/08 04:25 PM

rev1527 - clean up remnants no longer used after switching to EJB3 time service for periodic, server-specific tasks;

[ Permlink | « Hide ]

Jeff Weiss - 29/Sep/08 03:02 PM

I saw this is fixed on rev1632.