ACE backup serverfarm failover question

Unanswered Question
Dec 13th, 2011

Hi there,

I have a scenario where a group of servers need to connect a pair of application servers, but only one app server can be active and have connections at a time.  I've setup a serverfarm with a single server as primary, then with a backup server that should go active if the primary server fails.  I've tested failover when the primary server fails, and the ACE successfully directs connections to the backup server.  My question is about the failover actions once the primary server comes back online.   I believe the default behavior is to maintain existing connections to the backup server, but start directing new connections to the restored primary server.  I've found that I can prevent the primary server from coming online by setting a long timer for the return to service, but my app developers are asking if the ACE can sever the connections to the backup server once the primary is restored.  Can anyone advise if this is possible?  So basically I want to kill any connections to the backup server in the event that the primary server becomes active again.

Here is my config:

probe tcp UAT_9781

  port 9781

  interval 10

  passdetect interval 5

  connection term forced

probe tcp UAT_9789

  port 9789

  interval 10

  passdetect interval 5

  connection term forced

serverfarm host ST-UAT

  rserver UAT-17 9781

    probe UAT_9781

    inservice standby

  rserver UAT-17 9789

    backup-rserver UAT-17 9781

    probe UAT_9789

    inservice

class-map match-all ST-UAT

  2 match virtual-address 10.7.0.102 tcp eq 9789

policy-map type loadbalance first-match ST-UAT

  class class-default

    serverfarm ST-UAT

    nat dynamic 2 vlan 2 serverfarm primary

policy-map multi-match INBOUND_POLICY

class ST-UAT

    loadbalance vip inservice

    loadbalance policy ST-UAT

    loadbalance vip icmp-reply active

Thanks for your help,

Brandon

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Marko Leopold Wed, 12/14/2011 - 00:49

You can try this

serverfarm host ST-UAT-PRIM

  rserver UAT-17 9781

    probe UAT_9781

    inservice standby

serverfarm host ST-UAT-BACK

rserver UAT-17 9789

    backup-rserver UAT-17 9781

    probe UAT_9789

    inservice

class-map match-all ST-UAT

  2 match virtual-address 10.7.0.102 tcp eq 9789

policy-map type loadbalance first-match ST-UAT

  class class-default

    serverfarm ST-UAT-PRIM backup ST-UAT-BACK

    nat dynamic 2 vlan 2 serverfarm primary

neovestit Thu, 12/15/2011 - 13:32

Hi Marko,

Thanks for the reply.

I tried your suggested config, but I noticed a couple of things:

First, configuring the backup-rserver under the ST-UAT-BACK serverfarm results in another rserver being created underneath that serverfarm. I don't know if that defeats the purpose of what you were trying to accomplish with your config.

Second, the policy-map config is making the standby rserver the primary serverfarm, which seems like it's backwards.  Testing this config out, I couldn't establish any connection until I reversed the order of the serverfarms in the policy-map.

Also, I've read through the configuration guides and I can't find anything that indicates connections to a serverfarm will ever be forcefully disconnected -- is that true?

Thanks,

Brandon

Marko Leopold Thu, 12/15/2011 - 23:36

Oh it was a copy&paste error. It should be like this:

serverfarm host ST-UAT-PRIM

  rserver UAT-17 9781

    probe UAT_9781

    inservice

serverfarm host ST-UAT-BACK

  rserver UAT-17 9789

    probe UAT_9789

    inservice

branfarm1 Fri, 12/16/2011 - 08:22

Hi Marko,

Unfortunately that didn't work either.   I just don't think this is going to be possible. From the documentation:

http://www.cisco.com/en/US/docs/interfaces_modules/services_modules/ace/vA2_3_0/configuration/slb/guide/classlb.html#wpxref39283

Enabling Load Balancing to a Server Farm

You can load balance a client request for content to a server farm by using the serverfarm command  in policy map load-balancing class configuration mode. Server farms are  groups of networked real servers that contain the same content and that  typically reside in the same physical location.

The syntax of this command is as follows:

serverfarm name1 [backup name2 [aggregate-state]]

The keywords, arguments, and options are as follows:

name1—Unique identifier of the server farm. Enter an unquoted text string with no spaces and a maximum of 64 alphanumeric characters.

backup name2—(Optional)  Designates an existing host (with valid content) or a redirect (sorry)  server farm as a backup server farm in case all the real servers in the  primary server farm become unavailable. You can configure one backup  server farm for each existing primary server farm. When at least one  server in the primary server farm becomes available again, the ACE sends  all new connections back to the primary server farm. The ACE allows  existing connections to the backup server farm to complete. You can  fine-tune the conditions under which the primary server farm fails over  and returns to service by configuring a partial server farm failover.  For details about partial server farm failover, see the "Configuring a Partial Server Farm Failover" section in Chapter 2, Configuring Real Servers and Server Farms.  Enter the name of an existing server farm that you want to specify as a  backup server farm as an unquoted text string with no spaces and a  maximum of 64 alphanumeric characters.


I also tried a sticky group, but that has the same basic functionality in regards to the backup:

http://www.cisco.com/en/US/docs/interfaces_modules/services_modules/ace/vA2_3_0/configuration/slb/guide/sticky.html#wp1137791

Backup Server Farm Behavior with Stickiness

When you associate a server farm with a sticky group using the serverfarm command in sticky configuration mode, the primary server farm inherits  the stickiness of the sticky group. The ACE sends requests from the same  client to the same server in the primary server farm based on the type  of stickiness that you configure in the sticky group. If you also  configure a backup server farm using the backup option of the same command, you can make the backup server farm sticky, too, by configuring the optional sticky keyword.

If all the servers in the primary server farm go down, the ACE sends all  new requests to the backup server farm. When the primary server farm  comes back up (at least one server becomes active):

If the sticky option is enabled, then:

All  new sticky connections that match existing sticky table entries for the  real servers in the backup server farm are stuck to the same real  servers in the backup server farm.

All  new non-sticky connections and those sticky connections that do not  have an entry in the sticky table are load balanced to the real servers  in the primary server farm.

If the sticky option is not enabled, then the ACE load balances all new connections to the real servers in the primary server farm.

Existing non-sticky connections to the servers in the backup server farm are allowed to complete in the backup server farm.


I was hoping there was a trick to getting it to work, but perhaps not this time.

Thanks again,

Brandon

Actions

Login or Register to take actions

This Discussion

Posted December 13, 2011 at 5:31 PM
Stats:
Replies:4 Overall Rating:
Views:1349 Votes:0
Shares:0
Tags: ace
+

Related Content