06-06-2012 02:49 AM
Hi everyone
We have a Problem with the default inactivity timeout values on our ACE A4.
A Web-Service of ours takes up to 12 Minutes for creating its reports. But after 300 secs, the ACE kills the connection and sends a Reset to Server and Client, due to Inactivity Timeout (see https://supportforums.cisco.com/thread/2140771).
As described in http://www.cisco.com/en/US/docs/app_ntwk_services/data_center_app_services/ace_appliances/vA3_1_0/configuration/security/guide/tcpipnrm.html, we configured the inactivity timer on a parameter-map, which worked out fine:
parameter-map type connection HTTP_TIMEOUT
set timeout inactivity 3600
policy-map multi-match CLIENT-VIPs
class ZONE1.XY.Z_CM
loadbalance vip inservice
loadbalance policy ZONE1.XY.Z_PM
loadbalance vip icmp-reply active
connection advanced-options HTTP_TIMEOUT
But this works only for Connections with Virtual IPs - Direct (bridged) connections as the previously mentioned service are not affected by this configuration. Is there any way to change the Inactivity Paramter in a global way and not only for VIP-Connections?
Thanks in advance and greetings from Berne, Switzerland
Stefan Mueller
Our configuration:
system: Version A4(2.2) [build 3.0(0)A4(2.2) adbuild_15:26:12-2011/10/10_/auto/adbure_nightly4/renumber/rel_a4_2_2_throttle/REL_3_0_0_A4_2_2]
system image file: (hd0,1)/c4710ace-t1k9-mz.A4_2_2.bin
Device Manager version 4.2 (0) 20110907:2229
Solved! Go to Solution.
06-06-2012 08:27 AM
Hi,
Maybe you can take a look of this feature:
Per CSCtf91257, the new switch-mode timeout command in configuration mode allows you to configure the inactivity timeout for TCP or UDP connections in Switch mode. The ACE forwards connections that do not match any VIP. In Switch mode, these connections have TCP normalization disabled and the inactivity timeout set to 2 hours and 15 minutes (8,100 seconds). Since UDP connections do not have a close protocol, this timeout defines their minimum lifetime. Therefore, this command was introduced to minimize the number of old connections, particularly UDP.
The syntax for this command is follows:
switch-mode timeout seconds
The seconds argument is the time period in seconds for idle connections after which the ACE disconnects the connection. Enter an integer from 1 to 65535. By default, the timeout is 8100 seconds.
For example, to configure a timeout of 10 seconds, enter the following command:
host/Admin(config)# switch-mode timeout 10
To reset the default timeout, enter the following command:
host/Admin(config)# no switch-mode timeout
Here you have the link where you can find extra details:
Hope this helps!!!
-------------------------------
Jorge
06-06-2012 03:22 AM
Hi,
Just a quick hint, create an ACL permiting our desire traffic, call that in a class map, call that class map in policy map and apply parameter map.
06-06-2012 08:27 AM
Hi,
Maybe you can take a look of this feature:
Per CSCtf91257, the new switch-mode timeout command in configuration mode allows you to configure the inactivity timeout for TCP or UDP connections in Switch mode. The ACE forwards connections that do not match any VIP. In Switch mode, these connections have TCP normalization disabled and the inactivity timeout set to 2 hours and 15 minutes (8,100 seconds). Since UDP connections do not have a close protocol, this timeout defines their minimum lifetime. Therefore, this command was introduced to minimize the number of old connections, particularly UDP.
The syntax for this command is follows:
switch-mode timeout seconds
The seconds argument is the time period in seconds for idle connections after which the ACE disconnects the connection. Enter an integer from 1 to 65535. By default, the timeout is 8100 seconds.
For example, to configure a timeout of 10 seconds, enter the following command:
host/Admin(config)# switch-mode timeout 10
To reset the default timeout, enter the following command:
host/Admin(config)# no switch-mode timeout
Here you have the link where you can find extra details:
Hope this helps!!!
-------------------------------
Jorge
06-15-2012 02:48 AM
thanks a lot, this helped
(although it isn't a global command: you have to config this for switch-mode and config advanced option for every VIP. A bit more of a hassle than we hoped for, but it works fine)
01-21-2013 02:23 AM
Hi,
We have had to change the http and https timeouts for all traffic and could not do it per VIP as we many times load balance tcp/any.
Had to open a TAC case and I think it was what Gaurav has suggested. This way all http & https traffic got a different inactivity timeout. But in another case that I just now have opened it looks like this is not true for connections that go through the alias IP, i.e. routed mode (more specific in our case One-Arm routed (asymmetric traffic)).
Here is the config we use on all our contexts where we need all HTTP/HTTPs to have 3600 as their inactivity timeout:
parameter-map type connection CONN-TIMEOUT-3600
set timeout inactivity 3600
class-map match-all CONN-TIMEOUT-HTTP
match port tcp eq http
class-map match-all CONN-TIMEOUT-HTTPS
match port tcp eq https
policy-map multi-match CONN-TIMEOUT-HTTP-HTTPS
class CONN-TIMEOUT-HTTP
connection advanced-options CONN-TIMEOUT-3600
class CONN-TIMEOUT-HTTPS
connection advanced-options CONN-TIMEOUT-3600
service-policy input CONN-TIMEOUT-HTTP-HTTPS
The multi-match policy was assigned globally and is an additional one the the one where we define the VIPs.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide