Logs show agent contacted, scheduler logs show nothing

I just want them to both run. It doesnt matter to me if they run at the sametime. The only reason i moved them to a minute or two apart is because i cant figure out why the agent doesnt show the master contacted it and the logs on the master do. I just thought maybe the agent couldnt handle two request at the exact sametime so i moved them

The debug log and scheduler log on both master and agent show nothing for the jib that runs on schedule. The monitor shows the job running but its not in the queue.

Ran by pushing run now buttom while job is selected
2 Wed Mar 26 14:41:04 EDT 2008 Wed Mar 26 14:41:04 EDT 2008 46 reporttable extract Remote job finished.
2 Wed Mar 26 14:40:36 EDT 2008 Wed Mar 26 14:40:36 EDT 2008 46 reporttable extract Agent "IAT" contacted.
2 Wed Mar 26 14:40:36 EDT 2008 Wed Mar 26 14:40:36 EDT 2008 46 reporttable extract Remote job started.

Ran by schedule
2 Wed Mar 26 14:01:07 EDT 2008 Wed Mar 26 14:01:07 EDT 2008 46 reporttable extract Agent "IAT" contacted.
2 Wed Mar 26 14:01:02 EDT 2008 Wed Mar 26 14:01:02 EDT 2008 46 reporttable extract Remote job started.

Ok. thanks. I guess I know now the "why" part.

The job is set to run detached and asynchronous. As a result, the scheduler spawns it as a separate process. The job fails on the remote side but never terminates locally. The solution, uncheck the detached property and you see some difference. Please let us know if that helps.

FYI: Script type and remote detached jobs don't notify the scheduler process when they fails. This is a design deficiency and it is about to change in the next build 4.1.254 which will be released next week.

How do i uncheck the detached option I can see its checked but its grey'd out and I cant change it.

Umm... good question. How to set a remote job not to be detached?

Here is the trick. Create new JavaScript script type job to be run once manually. Make sure this job is not set detached. This is doable in the Wizard. for the script enter

Scheduler.setJobProperty(777, "DETACHED", "N")

Replace 777 with the job id of the remote job. Close wizard and run the script job. you can delete the script job after the run..

If that doesn't help, your only other easy option is to wait for the next build. More complicated solutions involve scripting for checking status of the remote job directly on the agent and returning it back to scheduler.

Ill give that a shot but am i understanding you correct. Turning the jobs on as asych and detached (detached is my only option) it might not report back to the master correctly? I changed all the jobs to run asynch and the program went ape crap on me. Matser says it started a bunch of jobs the agents scheduler log never even sees that the master contacted it.

Simple answer to that, it is recommended that jobs are always run detached and synchronous. Exceptions can be made for jobs that take veeeeeery long to run (read this as hours or days) and you don't want them to block their queues. Job queues should be used to control job concurrency and sequencing and organize jobs for different applications, basically create and control job streams. Typically running 5 to 10 job queues provides optimal solution for most environments. On the other hand, detached mode protects the scheduler from offending jobs crashing the entire system in case something goes badly.

Hope this helps.

I understand what you are getting at and have worked with mainframes and understand the job flow you are talking about. The problem with the queues is if a job gets stuck the everything behind it doesnt run. There is so many ways do things in your software I am sure i can find away to work around it I just dont understand why running jobs asynch returns such bizzare results. Reading the users guide and your answers i cannot see why i would want to uncheck the detached mode anyways... again just strange the option is there but cannot be used. Thanks you are always quick to try to help.

I wish I could help more efficiently. It is just sometimes troubleshooting stuff you cannot see is very difficult and requires lots of ifs and tries in order to figure out the root cause of a problem. or at least find out how to avoid it.

Async. property can be used, but it makes jobs compete for internal and external resources and allows them to step on each other.

Any idea what would cause asynch jobs to not work at all? I can check the asynch box and it doesnt work with the agent. logs show agent was contacted the but the scheduler.log file on the agent never shows anything. I uncheck asynch and the job runs fine. I have several jobs that need to run asynch and dont want to make 7 sperate queues for these specific jobs. I am running the jobs as differnent user , remotely.

Let's take a look at the debug log and find out if the job gets aborted prematurely.

I think it is a good idea to isolate one job, and troubleshoot it separately while other jobs are not running. In order not to stop everything, you can copy the entire scheduler directory to a different place and there start a second 24x7 instance and delete all jobs, but one you want to troubleshoot first. This way, the local to that instance debug.log will only have stuff related to that particular job and no other "noise"

Ok this job runs every hour asynch. Here is the latest scheduler.log file from the agent.

31-Mar-2008 10:06:56 AM 2 null 46 reporttable extract Remote job started.
31-Mar-2008 10:06:56 AM 2 null 46 reporttable extract Job started.
31-Mar-2008 10:07:28 AM 2 null 46 reporttable extract Job finished.
31-Mar-2008 10:07:28 AM 2 null 46 reporttable extract Remote job finished.

*** as you will see it never shows the 11:05 job starting***

Here is the debug log from the agent the only thing it shows is me trying to kill the Job from the master and it cant find it.

2008-03-31 11:38:30,694 [RMI TCP Connection(3280)-192.168.171.45] ERROR com.softtreetech.jscheduler.business.runner.JobExecutorImpl - killJob: unable to find job with run-time #2285

Neither the scheduler or debug log on the Master scheduler have anything at all but, if you look at the Logs in the gui it says this and I see the job running in the background queue and in job monitor

31-Mar-2008 11:05:15 datablaster-at 46 reporttable extract Agent "IAT" contacted.
31-Mar-2008 11:05:07 datablaster-at 46 reporttable extract Start message sent.
31-Mar-2008 11:05:00 datablaster-at 46 reporttable extract Remote job started.

Which build are you running on the master and agent? You can see that in the Help/About box the last 3 digit part in the version number.

I'm asking for that, because there were 1 or 2 relatively recent builds that had a problem with master scheduler overwriting some records in log files as a result of a known JRE bug. Sun says that it only occurs on Windows XP, but we were also able to reproduce it on Linux and Win 2003. The mentioned bug incorrectly positions file pointer, not at the end of the log file, effectively causing some overflow.

PS. We have a fix working around that JRE bug implemented in the last 2 builds. If this is a root cause of the issue, I suggest upgrading to 24x7 build 4.1.255 which will be available in a couple of days.

Multi platform edition Version 4.1 Build 247 is exactly what it says. Do I need to upgrade?

It might be one of these, but I need to find the notes and in the bug tracker database to say for sure.