cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
1503
Views
5
Helpful
3
Replies

PPPoE Session Stuck on Abnormal Termination [ASR 9010]

Asad Ul Islam
Level 1
Level 1

 

i am facing an issue here with BNG on ASR 9010 with IOS XR 4.3.1 

Whenever subscriber faces an abnormal/non-graceful termination like incase of modem power failure or cable unplug.  Their PPPoE sessions remain connected for nearly 3 minutes even after disconnection.

following is the configuration...... can someone tell me what can be done to resolve this issue??

 

dynamic-template
 type ppp PPP_TPL
  ppp authentication pap chap
  keepalive 30
  ppp ipcp dns 8.8.8.8
  ppp ipcp peer-address pool test
  accounting aaa list default type session periodic-interval 30
  ipv4 verify unicast source reachable-via rx
  ipv4 unnumbered Loopback0
 !
 type ppp PPP_NO_LB
  ppp authentication pap chap
 !
!
pppoe bba-group global
 service selection disable
!
class-map type control subscriber match-any PPP
 match protocol ppp 
 end-class-map
!
!
policy-map type control subscriber PPP_PM
 event session-start match-first
  class type control subscriber PPP do-until-failure
   10 activate dynamic-template PPP_NO_LB
  !
 !
 event session-activate match-first
  class type control subscriber PPP do-until-failure
   10 authenticate aaa list default
   20 activate dynamic-template PPP_TPL
  !
 !
 end-policy-map
!

 

1 Accepted Solution

Accepted Solutions

with the scale that the a9k supports (256k today, 512k tomorrow) having a keepalive value of 1 second will drive the cpu insane.

for safety reasons, the minimum value is set to 30 seconds.

In IOS for at that time the high scaling platforms like c10k, we always recommended interval of 60 seconds to ease the cpu.

If that takes too long for your taste you can set the retries lower, to say 3, but I wouldnt go any lower then that to prevent "false positives" and creating too much session churn.

XR 51 has the ability to set it to 10 seconds minimum but I would NOT recommend that:

RP/0/RSP0/CPU0:A9K-BNG(config)#dynamic-template type ppp TEST
RP/0/RSP0/CPU0:A9K-BNG(config-dynamic-template-type)#keep ?
  <10-180>  Enable keepalives with this interval (default is 60 seconds for RP S
ubscriber, 120 seconds for LC Subscriber)

xander

View solution in original post

3 Replies 3

Asad Ul Islam
Level 1
Level 1

OK i Got it

PPP brings down session on 5 keep alives misses. & since default keep alive is 60 seconds.. it takes 300 seconds before ASR will flush the session. 

Now the problem is that the minimum keepalive we can set is 30 seconds in XR, where as in IOS it could be from 1 seconds onwards.

Is there anyway to set it lower than 30 seconds?? how about in 5.x version?

with the scale that the a9k supports (256k today, 512k tomorrow) having a keepalive value of 1 second will drive the cpu insane.

for safety reasons, the minimum value is set to 30 seconds.

In IOS for at that time the high scaling platforms like c10k, we always recommended interval of 60 seconds to ease the cpu.

If that takes too long for your taste you can set the retries lower, to say 3, but I wouldnt go any lower then that to prevent "false positives" and creating too much session churn.

XR 51 has the ability to set it to 10 seconds minimum but I would NOT recommend that:

RP/0/RSP0/CPU0:A9K-BNG(config)#dynamic-template type ppp TEST
RP/0/RSP0/CPU0:A9K-BNG(config-dynamic-template-type)#keep ?
  <10-180>  Enable keepalives with this interval (default is 60 seconds for RP S
ubscriber, 120 seconds for LC Subscriber)

xander

Yes that makes sense.

We will reduce the retries to compensate the delay. 

Thank you :)