Multiple Jenkins Instances, Single CIFS file system.

[Eric Krystof]

Discussing an interesting scenario with my tech lead at work today.  Might try this out but was looking for any definitive remarks indicating we may be trying to divide by zero.

We need to build up a few slaves since our single instance is becoming overwhelmed.  The thought was that since this is entirely local network-wise that we possibly try this:

Master system with all jobs and workspaces in directory X.

Slaves setup to mount X as CIFS to run their jobs on.

We need the horsepower, not the space, so we're trying not to waste it, hence the CIFS idea (which could add to build time, which could be acceptable).

So I'm wondering... is that too far off the rails?  Would we need to segregate jobs to run only on certain slaves?  Does the shared mount bring along issues of configurations being overwritten or not read properly by each Jenkins instance?

Thanks for pondering.

[Mark Waite]

I think it could be made to work with enough exploring and experimenting, but it seems to me like you're using something expensive (human time, thought, planning, diagnosis) in order to conserve something that is relatively inexpensive (hard drive capacity).  Why not just buy more hard disc space, avoid the complexity and reduce the network traffic?

I think you're probably also sacrificing execution time in your desire to save disc space.  In my case, execution time is the most critical aspect of the Jenkins jobs, because we can only provide feedback to users as fast as we can cycle through the jobs.

I'm sure there are cases where the trade-offs you're considering are exactly the right trade-offs (for example, massive video production where you cannot afford to have more than a single copy of the workspace, a monster finite element analysis job where a single set of working files may overwhelm your capacity, etc.).

Mark Waite

---

From: Eric Krystof <erkr...@gmail.com>
To: jenkins...@googlegroups.com
Sent: Friday, May 17, 2013 7:07 PM
Subject: Multiple Jenkins Instances, Single CIFS file system.

--
You received this message because you are subscribed to the Google Groups "Jenkins Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-use...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Mandeville, Rob]

The other question is, what sort of horsepower do you need?  Are you CPU-bound or I/O bound?  The setup that you describe will allow you to leverage the CPU and memory of the slave systems, but if you’re I/O bound, your jobs will all be waiting on the same disk platters.  I once had (not even using Jenkins) four build and test servers all using the same NFS directory (NFS like CIFS for Unix).  The build servers were fine, but the NFS server was brought to its knees.

 

Your directory X scares me, or maybe I’m not reading it right.  Each slave has its own unique “remote FS root”, so anything being done there will be isolated from work on the other slaves.  By directory X, do you mean that you want the master and slaves to all work on the same filesystem, or literally the same directories?  If they’re all going to be working on the same directories, then the jobs will interfere with each other and you’ll have race conditions.  If you mean that they’re all using different directories in the same filesystem, then just worry about being I/O bound.

 

--Rob

The information in this message is for the intended recipient(s) only and may be the proprietary and/or confidential property of Litle & Co., LLC, and thus protected from disclosure. If you are not the intended recipient(s), or an employee or agent responsible for delivering this message to the intended recipient, you are hereby notified that any use, dissemination, distribution or copying of this communication is prohibited. If you have received this communication in error, please notify Litle & Co. immediately by replying to this message and then promptly deleting it and your reply permanently from your computer.

[Eric Krystof]

Mark, Rob, thanks for the insight.

It was all about the disk space, and for us it's not as cheap as we'd like.  We were able to procure a few extra hundred GB which helps.  It was more of a thought in case we couldn't.  Nevertheless, thanks for your replies, and, based on your responses, I'm glad we no longer need to go down this road as we'd run into the same issues you described.

Cheers,

Eric R. Krystof