02-03-2010 07:58 AM
Hello,
Just wondering why after a WAAS reboot, the first WAE devices that was normally handling the bulk of the optimization traffic is now sending packets to the second WAE device? Shouldn't WCCP be smart enough to know this is the lead WAE and to continue optimizing?
show wccp wide-area-engine
Wide Area Engine List for Service: TCP Promiscuous 61
Number of WAE's in the Cache farm: 2
Last Received Assignment Key IP address: x.x.x.68
Last Received Assignment Key Change Number: 11
Last WAE Change Number: 5
Assignment Made Flag = FALSE
IP address = x.x.x.69 Lead WAE = NO Weight = 0
Routers seeing this Wide Area Engine(2)
x.x.x.193
x.x.x.197
IP address = x.x.x.68 Lead WAE = YES Weight = 0
Routers seeing this Wide Area Engine(2)
x.x.x.193
x.x.x.197
Wide Area Engine List for Service: TCP Promiscuous 62
Number of WAE's in the Cache farm: 2
Last Received Assignment Key IP address: x.x.x.68
Last Received Assignment Key Change Number: 11
Last WAE Change Number: 5
Assignment Made Flag = FALSE
IP address = x.x.x.69 Lead WAE = NO Weight = 0
Routers seeing this Wide Area Engine(2)
x.x.x.193
x.x.x.197
IP address = x.x.x.68 Lead WAE = YES Weight = 0
Routers seeing this Wide Area Engine(2)
x.x.x.193
x.x.x.197
show wccp gre
Transparent GRE packets received: 557
Transparent non-GRE packets received: 0
Transparent non-GRE non-WCCP packets received: 0
Total packets accepted: 291
Invalid packets received: 0
Packets received with invalid service: 0
Packets received on a disabled service: 0
Packets received too small: 0
Packets dropped due to zero TTL: 0
Packets dropped due to bad buckets: 0
Packets dropped due to no redirect address: 0
Packets dropped due to loopback redirect: 0
Pass-through pkts dropped on assignment update:0
Connections bypassed due to load: 0
Packets sent back to router: 0
GRE packets sent to router (not bypass): 516
Packets sent to another WAE: 71
02-03-2010 11:49 AM
There is a feature called flow protection within WCCP to maintain state of the TCP connection during a WAE insertion.
Thanks
Eric
02-03-2010 12:06 PM
Thanks for the information. The WCCP flow protection is turned on by default. If I were to disable it, will the lead WAE resume optimizing the connections all the time after a reboot?
02-03-2010 12:14 PM
WCCP hashing/masking is based on the source IP or destination IP. If you have 2 WAE devices the WAEs will spilt this bucket assignments and pass some traffic to WAE A or WAE B based on the bucket assignment.
If you disable this feature then the client with get a packet that it was not expecting and and will cause a TCP reset. Then the client will have to re-establish the TCP connection.
Thanks
02-03-2010 12:46 PM
ok I understand the WCCP hashing/mask and the bucket assignments, with HASH you only get 50% and you can't carve it up to get more granular. The question I have is when a WAE is rebooted and returns to participate in WCCP, why is 100% of the traffic being redirected to the rebooted WAE then being sent to the other WAE? I would like is so when the rebooted WAE is back online it receives the original connections without sending it to the other WAE. It's and extra hop that doesn't seem right, all traffic is sending to the rebooted WAE but then turns around and sends it out to the other WAE for optimization. If it's maintaining existing connections that I can understand, but for new connections its establishing on the other WAE, not the rebooted one. The only way I figured around this is de registering/re registering WCCP at the same time on both WAEs.
So by disabling WCCP flow control help in this situation?
02-03-2010 05:57 PM
When a WAE reboots and reconnect to the WCCP group, it will start receiving traffic once it is registered based on the bucket assignment. However since the TCP traffic is still established the goal is to mininize additional outage to the end user. WCCP flow protection will redirect the traffic of established connections back to the WAE that is handling the traffic. Once new TCP connections are seen the new WAE should start to optimize that traffic. Are you seeing this?
Regarding disabling flow protection, I am not sure since i never really disable this feature to comment.
Thanks.
02-04-2010 07:49 AM
No, any new traffic is going to the rebooted WAE and then being sent to the second WAE for optimization. The hitcoutns are incrementing a lot for "Packets sent to another WAE". I've just RDP to a server, it went through the rebooted WAE and is being optimized on the second WAE. So, the only way to correct this is to stop WCCP on both boxes and reenabled it at the same time.
02-09-2010 08:50 AM
Any idea to why I'm still getting hit counts on this? Is the only way to resolve this is to remove WCCP and turn it back on?
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: