Monday, November 27, 2017

Self Healing Fabric with OCUM & WFA

Who doesn't love the concept of a self-healing system.  But what is self-healing ?  How can it be accomplished.  And can I make it fit my needs ?

In this "cookbook"-tutorial, you can find at least 1 method how it can be done.  But once you know the concept, the possibilities are nearly endless.

This idea is originally from Yaron Regev, so all creds go to him :).


Below the youtube link.  It starts with the concepts first, you can fast-forward to the demo's which are pretty cool (as a concept at least).

You can also download this zip.
It contains the 2 sample scripts, a backup of my WFA & my lab creation script, in case you would like to simulate it yourself.

Preparations to simulate this tutorial :

- Install WFA 4.1
- Restore the backup
- Install the Cdot pack 1.3.2
- Change the datasource to your OCUM
- Change the credentials for your clusters
- Create a database-report user in your OCUM
- Change the lab-script to fit your needs.
- My labscript has a dependency to my logger module, which you can find here.
- My labscript has a dependency to the manageability SDK, which you can download from the support-site.  It's needed to refresh OCUM from powershell.
- My labscript copies some larger files to my scenario volume to it make it grow from 70% to 90%.  You'll have to find your own sample files to play with.  Just some images or installer files will do fine.

on request: the dar file to modify the inode count :  download dar

8 comments :

  1. Hi Mirko, at the 29:20 mark you mention a command that already exists in WFA called "Modify Volume Max Inode Count". This same command was referred to in the session in Las Vegas. I can't find this command in my WFA implementation. I'm running 4.1 and have downloaded the latest .dar for Managing cluster mode but no luck. Is this a command that you created or is there a download somewhere to get access to it? Thanks for any assistance you can provide!

    ReplyDelete
  2. It's a custom command in the backup. I'll strip the command as a dar. Put it as a download on this post.

    ReplyDelete
  3. Hi,

    can you please provide me your email in order to send you some question?

    i will appreciate that i want to know a few automation maybe you have some ready.

    thanks,
    Gary Katz

    ReplyDelete
  4. Replies
    1. Hi,
      i have downloaded the inodes dar that you created and i want to know if you have a way to connect it to an alert that comes out of the ocum?
      i would like that every volume that will get to the 90% inodes full it sends out an alert to the ocum and i want that this script will kick in and clear the alert with out me knowing about it.
      i am running OCUM on the linux version that means i need some linux script that will start the wfa flow of the inodes increase.

      thanks,
      Gary Katz

      Delete
  5. Hi, yes that would be possible. But you are right, you would need some python/perl script to capture the arguments from ocum and trigger the workflow. My blog has a perl script to start workflows (but that's not mine). But you would have to extend it so it would
    - capture and group the arguments so you can filter out the sourceid, eventname and eventtype
    - filter only the events for inodes
    - filter only the new events
    - do a mysql lookup of the source id, finding volume, vserver & cluster name
    - trigger the workflow, passing volume, vserver & cluster.

    I have the script available in powershell (it's in the zip file provided above).
    I also rewrote the script once so the script fixes the inode problem directly, bypassing the wfa server.

    You can direct email me at mirko [at] netapp [dot] com

    ReplyDelete
    Replies
    1. okay thank you very much for your answer i will check out some perl script and if i will need some help i will contact you.
      i appriciate your help

      Gary Katz

      Delete