Press "Enter" to skip to content

vTIPS # 1

Hey all,

This is the first of my vTIPS series of posts, These are basically a quick tip that may come in handy when working with your virtual environments.

Load balancing monitors for products like vCAC, vCD, basically any product that can be load balanced.

When load balancing using the normal out of the box monitors that come with in this example F5 which is the more widely used HW load balancer just do not cut it.
For example when vCAC is sitting behind a load balancer the service may die but the tomcat service is still working fine so the normal health monitor will show as up and users are directed to a node that is down.
What we want is a custom monitor that will show the node as down if we get a 50x http error response.

In the following example we will be working with F5 virtual appliance 11.6 and vCAC 6.x
Send String: GET /vcops-web-ent/login.action HTTP/1.0\r\n\r\n

because we need SSL we require a double up of the r and n. we also specify the login page this will give us a http 200 success response. if we left the login.action off the response will give a redirect (30x) http response followed by a 200 response and will work with the receive string below as well.

Receive String: HTTP/1.[01] [23]0[0-6]

This is a regular expression that say we expect to see a 20x or 30x
Screen Shot 2014-09-22 at 9.51.05 am

Done add this to your pool and you will now know when a node is actually down.

Cheers

2 Comments

  1. Christopher Newell
    Christopher Newell September 25, 2014

    Hi Scott,
    Thanks for the vTip. We recently had this issue when deploying vCAC and required some investigation to get the correct send and receive string. This post would have come in handy.

    For a future tip, i would love to find a way to empty the vCAC database/s of all provisioned VMs. We wish to use vCAC as a provisioning tool only and perform all management through vSphere. The easiest way i can see to ensure there is no restrictions on vSphere management (for changes such as moving clusters between vCenters) is the empty the DB on a regular basis.

    • Scott Norris
      Scott Norris September 26, 2014

      Hi Chris,

      Not a problem Im happy to see people find some of my posts slightly usefull 🙂 . Thank you very much for your response.

      That is a great Idea for another vTIP. its unfortunate the removal of stale entries in vCAC is not a straight forward removal. I have seen lots of organizations using it purely as a provisioning tool so this has come up many times.
      It requires the removal of the entry from both the iaas SQL database and the postgres cafe database. I will have a look at wrapping that up in vCO or a script or something to make it user friendly.
      Ill have a look.

      cheers

      Scott

Leave a Reply

Your email address will not be published. Required fields are marked *

Anti SPAM BOT Question * Time limit is exhausted. Please reload CAPTCHA.