CSS 11501 service redirect

Answered Question
Aug 26th, 2008
User Badges:

Hello


I am new to the CSS 11501 content switch. I have it working with 2 virtual webservers...however when the web admin does maintenance on one web server they shut down apache and leave the virtual ip up...the content switch still sends requests to the server with the apache service down....can i configure it to check a port and if it is down sned the request to the other server?


thanks


Steve

Correct Answer by branfarm1 about 8 years 11 months ago

Glad it to hear works. The CSS is a pretty powerful platform and it can be configured in a lot of different ways. Be sure and check out the docs for other features you might be interested in.


Please rate if you found my comments helpful.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (2 ratings)
Loading.
branfarm1 Tue, 08/26/2008 - 11:02
User Badges:
  • Bronze, 100 points or more

Steve,


Yes you can configure it to check a specific port. When you configure the service, add the commands: keepalive port xx, and keepalive type http. I'm not sure if you have to specify port 80 if you use an http keepalive type, but I'm sure it can't hurt.


Be sure and test this before you roll it to production -- stop apache on one of the webservers and do show service summary. The service should show as 'DOWN' after the keepalive timeout expires.


--Brandon

sluciano Wed, 08/27/2008 - 02:14
User Badges:

Brandon


thanks for the response. i will give that a try and let you know how i make out.


Steve

branfarm1 Wed, 08/27/2008 - 06:11
User Badges:
  • Bronze, 100 points or more

Ajaz/Steve,


The sorry-server feature is used when you only want to have one active server at a time. The sorry server is configured as a backup for when the primary server isn't available. When load-balancing between two or more servers, the CSS will automatically stop sending requests to the down server, as long as it knows the server is down. If you want to have only one server active at a time, the sorry-server is definitely the feature to use, as Ajaz indicated. However, if you want to load-balance between 2 servers at the same time, then you need to work with the keepalives and the keepalive port to ensure that the CSS see's the service as down.


--Brandon

sluciano Wed, 08/27/2008 - 06:15
User Badges:

Brandon/Ajaz


i definitely want to load balance between the two...i have it configured with the keepalives....i'm just waiting for the web admin to shut down the port.....i'll reply as soon as i can test it.


thanks

Steve

sluciano Wed, 08/27/2008 - 06:38
User Badges:

we shut the services down on WEBSERVER1 and the content switch continued to send requests to WEBSERVER1 on port 80. does the content switch send all requests to one server based on a session or individual requests.....here is part of the config..


!************************** SERVICE **************************

service WEBSERVER1

ip address 10.15.1.11

keepalive frequency 2

keepalive port 80

active


service WEBSERVER2

ip address 10.15.1.12

keepalive frequency 2

keepalive port 80

active

branfarm1 Wed, 08/27/2008 - 06:43
User Badges:
  • Bronze, 100 points or more

Steve,


Can you post a full config? Also, can you post the outputs of show service summary from when Webserver1 is shutdown, and the output of "show service WEBSERVER1?


Thanks

sluciano Wed, 08/27/2008 - 06:47
User Badges:

here is the sh run....and the sh service summary


CSS11501# sh service summary


Service Name State Conn Weight Avg State

Load Transitions


WEBSERVER1 Alive 0 1 2 6

WEBSERVER2 Alive 0 1 2 22

WEBSERVERUK1 Alive 0 1 2 0

WEBSERVERUK2 Alive 0 1 2 0




Attachment: 
branfarm1 Wed, 08/27/2008 - 06:51
User Badges:
  • Bronze, 100 points or more

Can you also post the output of "show service WEBSERVER1" and "show service WEBSERVER2"?

sluciano Wed, 08/27/2008 - 06:52
User Badges:

here you go...

CSS11501# sh service WEBSERVER1


Name: WEBSERVER1 Index: 1

Type: Local State: Alive

Rule ( 10.15.1.11 ANY ANY )

Session Redundancy: Disabled

Redirect Domain:

Redirect String:

Keepalive: (ICMP 2 3 5 )

Keepalive Encryption: Disabled

Last Clearing of Stats Counters: 08/27/2008 03:06:52

Mtu: 1500 State Transitions: 6

Total Local Connections: 50036 Total Backup Connections: 0

Current Local Connections: 0 Current Backup Connections: 0

Total Connections: 50036 Max Connections: 65534

Total Reused Conns: 0

Weight: 1 Load: 2

Weight Reporting: None


CSS11501# sh service WEBSERVER2


Name: WEBSERVER2 Index: 2

Type: Local State: Alive

Rule ( 10.15.1.12 ANY ANY )

Session Redundancy: Disabled

Redirect Domain:

Redirect String:

Keepalive: (ICMP 2 3 5 )

Keepalive Encryption: Disabled

Last Clearing of Stats Counters: 08/27/2008 03:07:16

Mtu: 1500 State Transitions: 22

Total Local Connections: 38022 Total Backup Connections: 0

Current Local Connections: 0 Current Backup Connections: 0

Total Connections: 38022 Max Connections: 65534

Total Reused Conns: 0

Weight: 1 Load: 2

Weight Reporting: None


CSS11501#

branfarm1 Wed, 08/27/2008 - 06:59
User Badges:
  • Bronze, 100 points or more

I can see in those outputs that the keepliave is still based on ICMP:


Keepalive: (ICMP 2 3 5 )


So the only way a service will go "DOWN" with this config is to shut down the box. You need to add a keepalive type to each service. Try adding 'keepalive type http' to each service and if that doesn't work, use 'keepalive type tcp'


--Brandon

sluciano Wed, 08/27/2008 - 07:09
User Badges:

we are still seeing the connections on WEBSERVER1.....type http didn't work so i tried type tcp


!************************** SERVICE **************************

service WEBSERVER1

ip address 10.15.1.11

keepalive frequency 2

keepalive port 80

keepalive type tcp

active


service WEBSERVER2

ip address 10.15.1.12

keepalive frequency 2

keepalive port 80

keepalive type tcp

active


branfarm1 Wed, 08/27/2008 - 07:11
User Badges:
  • Bronze, 100 points or more

What does the 'show service WEBSERVER1' show?

sluciano Wed, 08/27/2008 - 07:12
User Badges:

CSS11501# sh service WEBSERVER1


Name: WEBSERVER1 Index: 1

Type: Local State: Down

Rule ( 10.15.1.11 ANY ANY )

Session Redundancy: Disabled

Redirect Domain:

Redirect String:

Keepalive: (TCP-80 2 3 5 )

Keepalive Encryption: Disabled

Last Clearing of Stats Counters: 08/27/2008 03:06:52

Mtu: 1500 State Transitions: 7

Total Local Connections: 50039 Total Backup Connections: 0

Current Local Connections: 0 Current Backup Connections: 0

Total Connections: 50039 Max Connections: 65534

Total Reused Conns: 0

Weight: 1 Load: 255

Weight Reporting: None


CSS11501#

branfarm1 Wed, 08/27/2008 - 07:13
User Badges:
  • Bronze, 100 points or more

That looks like it should be working now. With the service as Down, it should not receive any new connections.

sluciano Wed, 08/27/2008 - 07:18
User Badges:

Brandon


hang tight....it may be working.....if i do a sh load i get 0 average response time for WEBSERVER1....is this a good indicator?


thanks


Steve

branfarm1 Wed, 08/27/2008 - 07:27
User Badges:
  • Bronze, 100 points or more

I don't think that shows what you want in this case. The average response in the show load command is telling you the average time it takes to create a new flow for that service, from flow setup to teardown.


I would use the show service summary, and show service WEBSERVER1 to see if the number of connections is increasing (Current Local connections)

sluciano Wed, 08/27/2008 - 07:35
User Badges:

Great.....it seems to be working...thank you for all the help...


Steve

Correct Answer
branfarm1 Wed, 08/27/2008 - 07:38
User Badges:
  • Bronze, 100 points or more

Glad it to hear works. The CSS is a pretty powerful platform and it can be configured in a lot of different ways. Be sure and check out the docs for other features you might be interested in.


Please rate if you found my comments helpful.

Actions

This Discussion