Was wondering if someone could point me in the right direction on what may be causing this. A few Unix jobs that when cancelled using the desktop client do not go to FAILED state. One of these jobs runs a script which goes out to a database to process messages. The cancel command seems to always be killing the script on the server, its just that the job doesn't show as FAILED and remains in EXEC state.
Anyone else run into this sort of issue?
Please open a Support case; on support.ca.com, for the behavior that you are experiencing so that we can review and address. If you do not have access to the support site, please contact the Service Desk at 1-800-225-5224 and a representative will be happy to create one for you. Please be sure to provide the following information:
Steven A. Bartolini
Sr. Support Engineer
Thanks Steven for the info. I have grabbed the entire log directory on the server the job starts at. Being that the issue occurred on Friday (6 days ago) I am probably guessing that it is too far gone?? I don't dive too deep into the logs so I am not sure if they will contain the info that is needed or if it is wiped off by now. Also, would I need to grab log files from the DE core server itself or are we just referring to the agent in which the job runs on? I can have the user of the jobs notify me when the issue occurs again and I can get more accurate logs if needed.
We will need just the agent logs for now unless we determine that there is an issue with your scheduling manager. If it occurred 6 days ago, it very well could be that those logs have rolled off by now. Just in case, however, go ahead and open the case and indicate in the case comments that you had this discussion on our communities page. In fact, include the URL for this discussion in the case comments so the Engineer has a point of reference. If we are unable to locate the information in the logs, we'll go ahead and close the case and wait for the issue to resurface. When it does resurface, you should open a new case and attach the logs to the case for our review. I hope this is helpful. Hopefully we will have some data to go off of your current logs.
Have a great day and an even better weekend!
Thanks Steven I opened up a case yesterday and have been contacted. Hope you have a great weekend ahead.
As per Steve, please open a support issue with CA Support. In the meantime, you can also test to see if the 'Cancel' works with a command or not?
You mentioned that script is not getting killed, the agent uses 'kill -9' to stop the process in its tracks. Please test this manually as well. See if you can manually run your script and then kill it.
You can also submit a job that runs 'sleep' command with argument of 30 or 60. Then try to 'Cancel' it.
Thanks again for contacting CA Communities.
Hi Nitin I mentioned that the script does seem to be getting killed. I have tested this on a few different agents, a few agents of which were having a problem, no issue occurred and the jobs went to FAILED state. As of now I cannot replicate the situation but the user of the jobs will be contacting me when the issue occurs again.