Event log errors

Ok, all jobs have been set to detached. I've run three of the job chains, and the popup event log error did come up with one of them; that job chain was not logged by 24x7.

Twelve 24x7.exe processes have been spawned and remain open. The job chains did not complete.

The event log shows the following error which is a first

Event Type: Warning
Event Source: Win32k
Event Category: None
Event ID: 243
Date: 17/07/2007
Time: 11:30:29 a.m.
User: N/A
Computer: KBDVFDSINT02
Description:
A desktop heap allocation failed.

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
Data:
0000: 00 00 00 00 01 00 56 00 ......V.
0008: 00 00 00 00 f3 00 00 80 ....ó..€
0010: 00 00 00 00 00 00 00 00 ........
0018: 00 00 00 00 00 00 00 00 ........
0020: 00 00 00 00 00 00 00 00 ........

Here is the event log error this time for reference
Event Type: Information
Event Source: Application Popup
Event Category: None
Event ID: 26
Date: 17/07/2007
Time: 11:30:29 a.m.
User: N/A
Computer: KBDVFDSINT02
Description:
Application popup: 24x7.exe - Application Error : The instruction at "0x039c704d" referenced memory at "0x03a01f20". The memory could not be "read".

Click on OK to terminate the program

For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.

I've just tried running through a chain in 24x7 process mode and the jobs fired until they got to the last one, at which point it should complete successfully. Instead it popped up the box that the event log has been reporting (with the same memory ref for each one), once for each step. I tried that twice and it failed both times. I've set the jobs back to attached and it runs through OK like that, although I suspect it will continue to fail intermittantly.

I've just rebooted the server to clear everything out, started 24x7 in process mode, run the chain in attached mode ok, changed it to detached and it fails.

I've just looked at another thread on these forums from Joshua Youngblood and seen some similarities between the two cases.

I don't know if your case is really similar or not as I don't really know what you are doing.

In Joshua's case, they have built nightly processes generating dynamic jobs, modifying, running and deleting them. The problems began when Joshua build another job to query statuses of dynamic jobs. Because of some clash in internal processing, dynamic job creation and status querying at the same time led to problems in the status querying job causing it to crash. In their case all jobs were set to run detached and the error didn't affect the entire system. To the best of my knowledge that issue has been resolved.

FYI: In their case we also couldn't reproduce the problem but we knew the exact place where the error was occurring, and the possible cause. In the end, some internal method has been re-implemented using a different technology.

The most difficult part always is to establish the error cause. Fixing the problem or finding a workaround is usually easy.

The reason I 'm seeing similarity is that I've noticed 24x7 reporting processes returning error code 3221225477, and the fact that we are both running the .27 patch.

If it's not the same issue, then what do you suggest? All detached jobs are failing.

There were 2 releases 3.4.27a and 3.4.27b within 1 day. So let's start here. Please download http://www.softtreetech.com/24x7/fixes/24x7.lib and copy it to the 24x7 directory overwriting the existing file. In case if you weren't lucky and grabbed 3.4.27a... this file update will patch your version to 3.4.27b. Restart the scheduler and check if you can reproduce the issue.

In brief, 3.4.27a contained several fixes including a fix for Joshua's problem related to job status querying and dynamic jobs. Yet this new release it introduced a brand new problem caused by new graphical menus. The problem in 3.4.27a was isolated to detached-asynchronous jobs. It was corrected on same day in the form of 3.4.27b build. The difference is really in just one 24x7.lib file, that's why we have posted it separately so people can quickly patch the installation without running full upgrade. The full upgrade is also available in the usual place.

Ok, i have applied that new lib file. I suspect I did have the .27a patch as detached jobs now function as a rule. I will keep running stuff through and see if that got rid of the other error as well.

Thank you for the status update. You should be able now to transition to all detached jobs. Right?
If it happens again, please make a note of which jobs were running just before and during the error. My goal is to help you with pinpointing the exact place of the error or specific conditions that trigger it.

All of the action jobs are now detached (including the asynch one)

The few maintenance jobs ie archive log files etc are still detached but I don't imagine it will make a difference. I should, by previous experience, have the failure again at least once within the next 24 hours.

Given that we are using jobrun to do job chaining would it be advisable to only make the first job of the chain detached so I don't get 5-10 24x7 instances running at the same time? If i'm not wrong that would achieve the same end of having the action run detached from the scheduler.

I don't think it will make any difference. JobRun will make all JobRun referenced jobs to run as part of the first job in the chain. So if the first job is detached, the rest of the jobs will run in the same detached process regardless of how they are setup.

Are you sure? I've observed a 5-step chain open 5 separate 24x7.exe processes in task manager then close them one by one. If jobrun overrode detached you wouldn't get that right?

I'm sorry, let me take it back. Yes you're correct the detached job property does affect child jobs. So it is sufficient to make only the first job in the chain to run detached; the rest of the chain should have the detached property unchecked.