cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
841
Views
0
Helpful
2
Replies

Where to apply probe

billmatthews
Level 1
Level 1

I've configured a simple HTTP expect probe

probe http 1

  description simple_http_test

  interval 2

  faildetect 2

  passdetect interval 2

  passdetect count 2

  receive 5

  expect status 1 555

  open 1

I then have a server farm with two real webservers.  My question is, where should I apply the probe?  I want it to test each real server and take the server out of service if the probe fails. I've highlted in red a few of the places I've considered (under each rserver, under the serverfarm, or under the rserver defined under the serverfarm).

rserver host test1
  ip address 10.10.10.1
  probe 1
  inservice
rserver host test2
  ip address 10.10.10.2
  probe 1
  inservice

serverfarm host test_servfarm
  probe 1
  rserver test1 80
    probe 1
    inservice
  rserver test2 80
    probe 1

    inservice

Thanks -Bill

1 Accepted Solution

Accepted Solutions

chrhiggi
Level 3
Level 3

Hello Bill,

The theory behind applying a probe at a global level is to catch a general failure of the server to remove it from all serverfarms it is configured under.  I.e. If the server crashed, an ICMP probe would be sufficient for catching a failure and setting the server to a non-operational state under all serverfarms.

EX:

rserver host test1

  ip address 10.10.10.1

  probe 1

  inservice

Applying it at a serverfarm level is standard.  At that level, every rserver configured under the serverfarm is tested using the probe in question.  Generally, you would have a probe that is specific to the service you are sending traffic to under that specific serverfarm.  If any of the servers were to fail the probe test, they would be removed from rotation only from the serverfarm where the probe failed. 

EX:

serverfarm host test_servfarm
  probe 1
  rserver test1 80
    inservice
  rserver test2 80
    inservice

Applying a probe under an rserver within a serverfarm is less commonly used.  You would generally only apply this where you need to test more than 1 type of service for the server in question.  Say you had IIS and MySQL running on 1 server in your serverfarm.  All other servers in the serverfarm only have IIS installed, however, all servers talk to the single server with MySQL.  For all servers, you would want to probe IIS, for the one specific server, you would want to also probe MySQL.  Hence, you would apply a probe at the serverfarm level for IIS and a single probe under the rserver for MySQL.

serverfarm host test_servfarm
  probe 1
  rserver test1 80

    inservice
  rserver test2 80
    probe 2

Regards,

Chris Higgins

View solution in original post

2 Replies 2

chrhiggi
Level 3
Level 3

Hello Bill,

The theory behind applying a probe at a global level is to catch a general failure of the server to remove it from all serverfarms it is configured under.  I.e. If the server crashed, an ICMP probe would be sufficient for catching a failure and setting the server to a non-operational state under all serverfarms.

EX:

rserver host test1

  ip address 10.10.10.1

  probe 1

  inservice

Applying it at a serverfarm level is standard.  At that level, every rserver configured under the serverfarm is tested using the probe in question.  Generally, you would have a probe that is specific to the service you are sending traffic to under that specific serverfarm.  If any of the servers were to fail the probe test, they would be removed from rotation only from the serverfarm where the probe failed. 

EX:

serverfarm host test_servfarm
  probe 1
  rserver test1 80
    inservice
  rserver test2 80
    inservice

Applying a probe under an rserver within a serverfarm is less commonly used.  You would generally only apply this where you need to test more than 1 type of service for the server in question.  Say you had IIS and MySQL running on 1 server in your serverfarm.  All other servers in the serverfarm only have IIS installed, however, all servers talk to the single server with MySQL.  For all servers, you would want to probe IIS, for the one specific server, you would want to also probe MySQL.  Hence, you would apply a probe at the serverfarm level for IIS and a single probe under the rserver for MySQL.

serverfarm host test_servfarm
  probe 1
  rserver test1 80

    inservice
  rserver test2 80
    probe 2

Regards,

Chris Higgins

That was a huge help, thank you!

Review Cisco Networking for a $25 gift card