limitation of active jobs?

Hi All,

Do you have any limitation of active jobs for stable system?
or TeamSite has it?

If you have any information,
share to me.

Thanks.

Find more posts tagged with

Comments

Adam Stoller

Please explain the question in more detail:

nipper

How fast can a car go ? Well that depends....

That being said, I usually see problems when people have > 10K jobs (active/hung/etc) on a server.

duric

First of all, Thank you for your answers.

Here are server information.
Version : 7.3.1 Build 20 Interwoven
OS : Red Hat Enterprise Linux Server release 5.9
Current jobs : about 4K.

Currently we doesn't have any issue, but I would like to know how you calculated it.

Thanks.

nipper

There is no limit, but there were a whole lot of laughs on the forum when someone complained about performance and idly mentioned they had 25K active jobs.

4K is too many. 40 (often) is too many. If you have that many only a handful are actively being worked on. The remainder are hung. You should clean them up and then determine why they are hung. Often either review tasks no one knew about, external tasks that failed, or CGI tasks which were cut short. Those fall into either training issues or WF bugs.

Adam Stoller

In addition to what Nipper said - you generally have to consider the number of tasks per job multiplied by the number of jobs - to get a sense of how big things are - and when you start to get "too many" performance tends to degrade (or at least it used to). The definition of "too many" was always a bit of a subjective thing - but I think Nipper's comments tend to be reasonably good on that score.

duric

Thanks your advice.

We have many branches so I can't check all hanging jobs and I have been cleaning old jobs per each month.
I just want know what limitation exists or how to get a way.

Good luck to all answerer.
Regards.

Adam Stoller

As stated, there is no hard-and-fast limitation.
The best thing to do is make sure your workflows have proper timeout handlers for any task that could get "stuck" - so that someone gets pinged to look into it and get them "unstuck".

Depending on the task / task-type - the "pinging" could be to the owner of the task or to one or more TS administrators.