Has anybody seen the issue that a scheduled job is executed concurrently from two servers? I think the job scheduler lock in the database should prevent such situations. Any ideas what might cause that?
First you need to check if the multicast is working as other BG should not kick off if the Beacon is broadcasting the message to other server.
Yes, the multicast is working. That's already been checked.
Could someone explain to me how the multicast and the prlock table are involved in scheduled job runs? What's the database lock for if it doesn't prevent the other BG from taking the same job?
Is your server(s) a Linux or Unix server? If so, check for Zombie or Ghost bg processes.
This problem is caused by a service not being completely killed when a bg services is stopped via the nsa or at the command prompt. To check to see if this is the problem do the following for each Clarity server in the cluster.
1. Run the following command at the command prompt:
Before V13niku status all
V13 and after
Service status all
2. Note how many bgs are running.
3. Now run this command:
ps -ef | grep java
You will see results that look something like this:
clarity1 20128 1 0 Apr14 ? 00:00:04
./nikubgcmd.conf wrapper.syslog.ident=b g
/bg.pid wrapper.name=bg wrapper.displayname=CA Clarity PPM Background
Server [bg ] wrapper.daemonize=TRUE
clarity1 20130 1 0 Apr14 ? 00:00:04
NOTE: There may be additional services showing as well.
4. Check to see if you have more bg services running than indicated in Step 1. If you do not have any extra bg services, this is not the problem. You will need to use debugs to chase this problem further.
1. If you have more than one bg service running on the server, kill the older process by running the following command:
kill -9 #####
kill -15 #####
Where ##### is the process id that you found when running the grep command in step 3. For example, if the process you want to kill has results that start like this:
20130 is the id for the process. So, this would be the command you would run:
kill -9 20130 OR kill -15 20130
2. If you are having difficulty determining which process(es) to kill, stop your bg service(s). Kill any remaining bg services per the above instructions.
NOTE: If you have more than one Clarity server in the cluster, this should be done for each Clarity server.
A ghost bg process would make sense but unfortunately it's a Windows server.
Retrieving data ...