Strange behavior with "run on all node in the group&

crazybiker · Post by **crazybiker** » 17 Jul 2013 9:57

hi everybody,

I have a little problem with controlm jobs.
i have few jobs (near 68 jobs) in a smart table linked together with conditions like below.

All these jobs must run separately (in sequential) on 60 servers (in parallele) so i've checked the box "run on all node in the group" on each job, but after newday processing i have this :

As treatment which run on a separate server, they musn't wait each other, so it must be something like thit:

Does someone know what's wrong and what shouls i do to correct it easily ? (i don't want to duplicate job by job,.....

)
(we are with controlm 6.4.1)

thanks in advance

Nightwatchrenband · Post by **Nightwatchrenband** » 18 Jul 2013 6:37

Are ALL of your conditions based on ODATE? I assume that after newday, you have a new ODATE for each set. If your conditions are set with this in mind, that should keep them separate by odate.

gglau · Post by **gglau** » 18 Jul 2013 10:09

"Run on all nodes" is just like ordering a job definition multiple times, one for each node in the group. The result as described is expected.

For the identical active jobs, the only difference is OrderID and NodeID. To connect jobs of the same node, condition name must be encoded with node ID, for example

<JOBNAME>-ENDED-%%NODEID

According to BMC documentation for condition name,

"system variables (but not other types of variables) can be specified as the entire value for this parameter"

It works but use at your own risk.

Your alternative is to collapse all the steps/jobs into only one job.

crazybiker · Post by **crazybiker** » 19 Jul 2013 12:06

hi,

Yes they are all on the same odate, and it can't be only one job.

The solution from gglau with the %%NODEID seems to work very fine

Thanks gglau for this solution
Thanks all for the answer