STUN

a.gooding · ‎05-19-2006

Hey,

having a little issue with a customer using STUN accross branches to communicated with the as400. configuration was done about two years ago, and the speed was great. lately they have been having issues with the speed.

more specifically. One of the remote branches had a power outage. when power came back up , they regained connectivity, but for some reason unknown to me, at that same time, all other branches reported slowness of the system.

Now, it eventually started to work properly after a few days and i thought that was that, but then the same branch had onother outage last week, and the same thing happened. the only difference now is that the speed hasnt improved.

we checked lines and i did a check on the serial interfaces etc, and im not seeig any errors what so ever, the speed from router to router is good, and i used extended ping of up to 3000 bytes and that was fine.

im stumped as to how i should go about troubleshooting the issue...

please help

aghaznavi · ‎05-25-2006

When the TCP stack has more than one packet in the retransmission

queue, it means that these packets were either dropped in the network or were

not reliably received by the remote end. Under this condition, TCP goes into

slow start, sends the first packet in the retransmission queue and backs off

the retransmission timer(KRTT). This repeated till we get an ACK for the

packet. No other packet in the retransmission queue is sent unless we get the

ACK for the earlier packet in the retransmission queue. When the ACK comes in,

TCP is supposed to retain the same KRTT value for sending the next packet in

the queue. However, IOS TCP backs off yet again, causing the packets on the

retransmit queue to be sent very slowly.

Once the packets in the retransmit queue are cleared (ACKed), TCP will return

to a normal state of sending data up to the window advertised by the peer.

a.gooding · ‎05-29-2006

alright, first question is huh?...hehe, kidding, that sounds exactly like what is going on. the strange thing however, is that we rebooted the controller again, and immediately everything went fine. network speed was back up.

in this instance, and from what im getting with you, can i limit the slow start to that specific branch , meaning, i can have this one branch cause network issues throughout all branches. what might be your solution.

thanks