Setup ACE probe for HTTP host headers for multiple sites on rservers

Unanswered Question
Jul 7th, 2008

We have multiple sites on each server. Is there a way to have the probe only stop traffic to a specific site or header and not take the complete rserver offline?

Thanks

Greg

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
Syed Iftekhar Ahmed Mon, 07/07/2008 - 20:36

If you are running multiple web servers on same servers using same IP addresse on the server. Then Host header field differentiate these web instances on the same physical machine.

Use header command under http probe definition to send appropriate HOST value with probe rquest

e.g

probe http site1

interval 2

faildetect 1000

passdetect interval 2

passdetect count 1

header Host header-value "www.site1.com"

expect status 200 200

probe http site2

interval 2

faildetect 1000

passdetect interval 2

passdetect count 1

header Host header-value "www.site2.com"

expect status 200 200

HTH

Syed iftekhar Ahmed

smith4108 Tue, 07/08/2008 - 06:33

Thank you, that worked. Still looking at what we had wrong in our original probes.

smith4108 Wed, 07/09/2008 - 07:21

I guess I am a little to quick on the response to say it is fixed. I didn't test properly. When we fail just a site the probe still takes down the whole server.

Any other ideas?

Here is what we put in:

probe http UAT-site-Http

interval 20

faildetect 3

passdetect interval 120

expect status 200 200

header Host header-value "www.site.com"

Syed Iftekhar Ahmed Wed, 07/09/2008 - 10:18

Is your web site listening on www.site.com?

If you type www.site.com on the browseryou get the page?

You are suppose to use multiple probes for multiple sites. Are you doing that.

What is the status of the other probes when you take one site down?

Syed

smith4108 Wed, 07/09/2008 - 10:25

yes the site still comes up on our browser, but the cookie shows us going to the other server all the time.

We are using a single probe for each web site.

The other probes show the other sites availble on the rserver.

We have tried using roundrobin and leastconnections. But as soon as one site is turned down all the traffic for all the other availble sites go to the one server with no sites down.

Greg

aylesworth Thu, 09/10/2015 - 15:48

Syed,

How are these applied to each serverfarm in multiples?

Currently we have one probe, SITECHECK applied to serverfarms via probe SITECHECK.

Can I assign multiple probes to the same serverfarm? Or do I assign the probes to the rservers themselves? If so, if one of the probes fails, will it still mark the rserver as down or will it still see it as up if the other probes still respond?

Thanks!

Syed Iftekhar Ahmed Thu, 09/10/2015 - 16:03

Multiple probes can be applied to a Serverfarm or Rservers and will have an OR logic applied by default (i.e any single failure will cause Rserver failure).

Its also possible to change his OR logic into AND logic using “fail-on-all" command

 

For e.g in the following case any probe failure will Mark the Rserver down

 

rserver SERVER1

   ip address x.x.x.x

   probe First_PROBE

   probe Second_PROBE

   

Whereas with  fail-on-all command , Rserver will be marked down only if both probes are down

 

rserver SERVER1

 ip address x.x.x.x

 probe First_PROBE

 probe Second_PROBE

 fail-on-all

aylesworth Thu, 09/10/2015 - 16:20

Syed,

First off, thanks for replying to a thread that's 7 years old :)  I hadn't noticed that.

That info is exactly what I needed, thank you for the advice!

aylesworth Thu, 09/10/2015 - 16:33

Also I have one more question: If one of the probes goes down and the fail-on-all flag is set, will the ace then forward traffic for only that failed probe to the second server in the farm?

Or will the probe-failed application still receive traffic, but the end users will receive an iis 500 or whatever the case is?

Syed Iftekhar Ahmed Thu, 09/10/2015 - 16:37

if fail-on-all is set and only one probe fails then ACE will not mark the Rserver as down.

Since Rserver is not marked down, all traffic will be sent to it as usual.

Users will experience the application error if they hit what is being served from the Application area which is not up. 

 

Actions

This Discussion