Have been happily using per-session QoS for a while now in 12.4M and am now forced to move to 12.2SR as part of a platform change.
Doing so requires a new RADIUS AVP implementation
(see
http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/per_session_qos.html
and
http://www.cisco.com/en/US/docs/ios/security/configuration/guide/sec_qos_aaa_policy.html
)
Which now seems to be broken as the "shape average percent XX" now seems to set the CIR
from the PHYSICAL interface the L2TP packets arrive on (gigE) as opposed to the virtual-access
which the subscriber arrives on, which has its bandwidth set correctly by AVP (a "sh int" confirms
that the vaccess bandwidth is indeed set correctly)
Disabling subinterfaces as a test does not fix this, s/w in question is 12.2(33)SRD, has anybody
come across this? bugtool yet again does not seem to yield anything useful.
Thanks in advance
Dave.
------------------------------------------------
David Freedman
Group Network Engineering
Claranet Limited
http://www.clara.net
Doing so requires a new RADIUS AVP implementation
(see
http://www.cisco.com/en/US/docs/ios/qos/configuration/guide/per_session_qos.html
and
http://www.cisco.com/en/US/docs/ios/security/configuration/guide/sec_qos_aaa_policy.html
)
Which now seems to be broken as the "shape average percent XX" now seems to set the CIR
from the PHYSICAL interface the L2TP packets arrive on (gigE) as opposed to the virtual-access
which the subscriber arrives on, which has its bandwidth set correctly by AVP (a "sh int" confirms
that the vaccess bandwidth is indeed set correctly)
Disabling subinterfaces as a test does not fix this, s/w in question is 12.2(33)SRD, has anybody
come across this? bugtool yet again does not seem to yield anything useful.
Thanks in advance
Dave.
------------------------------------------------
David Freedman
Group Network Engineering
Claranet Limited
http://www.clara.net