TCP connection inactivity timeout

[shday]

Is it true that the ACE module has an tcp inactivity timeout of 3600 seconds (1 hour)?

[Cesar Roque · Accepted solution]

Yes, it is true.  These are the default values:

Parameter-map : new
Description : -
Type : connection
   nagle                              : disabled
   slow start                         : disabled
   buffer-share size                  : 32768
   inactivity timeout (seconds)       : TCP: 3600, UDP: 120, ICMP: 2===================HERE IT IS
   embryonic timeout (seconds)        : 5
   ack-delay (milliseconds)           : 200
   WAN Optimization RTT (milliseconds): 65535
   half-closed timeout (seconds)      : 3600
   TOS rewrite                        : disabled
   syn retry count                    : 4
   TCP MSS min                        : 0
   TCP MSS max                        : 1460
   tcp-options drop range             : 0-0
   tcp-options allow range            : 0-0
   tcp-options clear range            : 1-255
   selective-ack                      : clear
   timestamp                          : clear
   window-scale                       : clear
   window-scale factor                : 0
   reserved-bits                      : allow
   random-seq-num                     : enabled
   SYN data                           : allow
   exceed-mss                         : drop
   urgent-flag                        : allow
   conn-rate-limit                    : disabled
   bandwidth-rate-limit               : disabled

[shday]

Do these timers still apply for traffic that is just being routed by the ACE and not load balancing the servers?

  I have a vlan that I moved from a Cat6500 to my ACE30 and for now I have no policies built I'm just using the ACE vlan interface as the default gateway for the servers on that vlan.  I have plans to build policies for this application but not just yet.  The issue is that the servers on the ACE are making a call to another set of servers not on the ACE and these particular connects (tcp session) must stay active all the time and if they go down we have to restart the services on the servers that are not on the ACE.  What we are seeing is that the sessions are being dropped after and hour presumably from the tcp 3600 second timeout.  I turned off normalization on the interface yesterday but we are still having the issue.  I had to move all the servers back the the 6500 and not allow these connections to go through the ACE.

I need a way to keep these tcp session active even if they become idle.

[Kanwaljeet Singh]

Hi,

"Inactivity timeout" value WILL affect connections "established through the ACE" even if they are NOT terminated on ACE or have a VIP configured for them.

Regards,

Kanwal

[Kanwaljeet Singh]

Hi,

You can also try and match the all traffic and create a pmap and apply to it. You can increase the inactivity time out as per your needs.

Here's a sample config:

parameter-map type connection TIME-OUT

  set timeout inactivity 30

access-list all line 8 extended permit ip any any

class-map match-all PASS

  2 match access-list all

policy-map multi-match LB

    class PASS

    connection advanced-options TIME-OUT

Apply this policy and check if that helps.

Regards,

Kanwal