Monitoring custom processes via scalr

24 views
Skip to first unread message

Srini

unread,
Mar 4, 2012, 4:45:22 AM3/4/12
to scalr-...@googlegroups.com
As previously mentioned in an earlier post, my app role actually runs Jboss. The Jboss instance could have crashed and become unresponsive. Given that it's an apache customized role, scalr is not even aware of this process.

Assuming that I can monitor jboss from the scalr server via shell scripts, what is the best way to integrate this into scalr. There are two requirements:

1. If the jboss app instances are all overloaded, a new instance should be spawned off based on some metric. I assume I will have to create a new scaling metric for this and provide it.

2. If the instance is not healthy, it has to be terminated and a new instance started. How do I tell scalr the instance is not healthy? Do I use the scalr command line tools and terminate the instance via the API? Or is there a better way to do that?

Any Help is appreciated.

Regards
Srini

Nick Toursky

unread,
Mar 4, 2012, 5:00:19 AM3/4/12
to scalr-...@googlegroups.com
Srini,

To scale jboss instance, I'd recommend to create a new scaling metric just as described in the following blog post: http://blog.scalr.net/feature/2-1-feature-highlight-custom-scaling-metrics/
If your script detected unhealthy instance, you can terminate it in any convenient way: by executing 'shutdown -h now' on it, by Scalr API or EC2 API. Second option is recommended, first is also good, the latter could result into elevated time to bring up a replacement.

Regards,
Nick

Srini

--
You received this message because you are subscribed to the Google Groups "scalr-discuss" group.
To view this discussion on the web visit https://groups.google.com/d/msg/scalr-discuss/-/RbBL1f876JoJ.
To post to this group, send email to scalr-...@googlegroups.com.
To unsubscribe from this group, send email to scalr-discus...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/scalr-discuss?hl=en.


Srinivasan S

unread,
Mar 4, 2012, 5:09:26 AM3/4/12
to scalr-...@googlegroups.com
Thanks Nick. I was not expecting an answer on Sunday and was planning to do this tmrw , now I can't stop myself from trying!

You guys rock.

Srini
Sent on my BlackBerry® from Vodafone
From: Nick Toursky <hin...@gmail.com>
Date: Sun, 4 Mar 2012 12:00:19 +0200
Subject: Re: Monitoring custom processes via scalr

Srinivasan S

unread,
Mar 4, 2012, 5:37:16 AM3/4/12
to scalr-...@googlegroups.com
Hi Nick

When you say second option is recommended do you mean scalr api? Which method would result in elevated time to bring up a replacement?

Before setting it up I just want to confirm the best possible way.

Cheers

Srini
Sent on my BlackBerry® from Vodafone
From: Nick Toursky <hin...@gmail.com>
Date: Sun, 4 Mar 2012 12:00:19 +0200
Subject: Re: Monitoring custom processes via scalr

Nick Toursky

unread,
Mar 4, 2012, 5:39:44 AM3/4/12
to scalr-...@googlegroups.com
Srini,

Using Scalr API is recommended. 
Using EC2 API could result into elevated reaction time.

Regards,
Nick

Srinivasan S

unread,
Mar 4, 2012, 6:17:27 AM3/4/12
to scalr-...@googlegroups.com
Ok got it.
Sent on my BlackBerry® from Vodafone
From: Nick Toursky <hin...@gmail.com>
Date: Sun, 4 Mar 2012 12:39:44 +0200
Reply all
Reply to author
Forward
0 new messages