DSXchange

Hi All,
I am able to open a job and I am able to change it but it is displaying the following error while compiling.....

"Can not get exclusive access to executable file for job may be being monitored"...

It is a multi instance job, and currently I have not opened any monitor in director......

Thanks....

Do you have this jobs log open in any director? Go to 'Cleanup Resources' under Job button in the director and see if you have any sort of log on any RT_LOG file.

I have checked it...but i do not have that RT_LOG file ya

Check if somebody in your team has got the director open and the current focus is on the said job(i.e if the said is job selected in the director at the moment).

If yes pls remove the focus and try again

there are many option to get rid of this

1. in DataStage Administrator choose correct project then go to Command and Enter DS.TOOLS. from the first menu use option "5" and from the second menu that appears choose "11"th option and then enter your job name correctly, rest DS Administrator will take care

2. use putty any ssh client to login to you Datastage home, make sure no one has any job running, then in type following command in given directory

DSHOME/bin/uv -admin -stop

wait for 30-40 secs and then enter

DSHOME/bin/uv -admin -start

it will restart the DataStage engine and all the jobs will be released.

3. Make sure there are no jobs running under your project,
Go to DataStage Adminstrator in your project fire the following command

CLEAR.FILE &PH&

this command will clear all phantom process in the project you fire this command and all the job will be released

4.Or if your job get hanged due to your database gone down while the job was running than in the UNIX login to your project and fire the following command

ps auxw | grep <project>

it will list all the process that are in your project find the pid of your job from the list and use-> kill -9 <pid> to kill the zombie entries this will also make your job free.

Thanks ya

After killing the job only I am getting the error like this ya...

A simple solution is to copy the design, paste it on a new canvas and save it under another name. This way the new job relieves itself of any locks , blah...blah, etc and should work fine. Of course, that is if you can afford to rename the job.

gateleys

Please do NOT advise kill until ALL other avenues have been exhausted.

The correct procedure is to identify the owner of the lock, with a view to getting it released cleanly. If it is a defunct process, the administrator has an UNLOCK tool available (perhaps via the DS.TOOLS menu).

A Director open in log view will have a lock on the RT_LOGnnnn hashed file. A Director open in monitor view will have a lock on the RT_CONFIGnnnn hashed file.

Killing processes should always be the very, very last resort, even after Clear Status File. And that's so last resort it asks "are you sure?" twice!

47shailesh - I have issues with several of your recommendations -

Recommendation 1 is a correct one.
Recommendation 2 is terrible, it will stop the DataStage engine - kick out any other users on the system who might be logged in and cause them to lose data if they were editing jobs and might leave running jobs in undefinaed states. In addition, unless all processes are actually logged out in UNIX the subsequent restart will not work.

Recommendation 3 is wrong, but won't harm the system like #2 can. The &PH& directory contains log files and clearing the contents can decrease some overhead involved in job administration, but no locks are held on these files and thus it doesn't pertain to the problem.

Recommendation 4 is DANGEROUS.
NEVER, EVER do a "kill -9" unless you have tried every other alternative (including a normal kill). I can guarantee that doing a "kill -9" will corrupt the instance so that you will have no option but to stop and start the engine. Those who know how to use the UNLOCK or ipcs/ipcrm commands can fix this without having to stop the system, but the same people would never think of using kill -9 in the first place.

47shailesh

I have cleared status file using ur option and logged out the specified job but again it is displaying error "Can not exclusive access, job can be monitored"............

Ray,
Where these files are available which you have mentioned........and how to unlock it
" RT_LOGnnnn hashed file. A Director open in monitor view will have a lock on the RT_CONFIGnnnn hashed file. "

Thanks for your reply....

Where they are is irrelevant - they are hashed files in your Repository.

Use list_readu command (or LIST.READU command or Cleanup Resources in Director) to see the locks. Use the DS.TOOLS menu or the UNLOCK command to free the locks held by defunct process(es).

Yeah like ray says, should be your best option! Believe me , I've done it and it works.
Use and to find the offending locks/processes. It will also give you the USERNO to unlock.

In DS Administrator use LIST.READU and see if you can find any locks related to your job. If you can find then note the pid, use DS.Tools command and choose option 4 and then option 7 and give the pid to release the locks. If that dosen't work then look for USERNO and use UNLOCK USER userno ALL. Make sure you are logged log in as a user with relevant privileges for this.