I have a customer with a server to server communication issue. Server A reports a .5k piece of data to Server B. It is supposed to retry 3 times if it fails. This has worked well for years for many customers however we have encountered one in which the reporting failures several times per day. After much troubleshooting and monitoring the network we have only identified one significant anomaly, the route from server A to B is changing on the customers side regularly.
It is 29 hops from server A to B and hops 10 to 28 change regularly, once every 5-10 minutes. When the entire route changes most of the hops change 4-6 times within 2 or 3 seconds and is then stable again for a while. Even with all this flapping about, I see no packet loss.
Question: Can this flapping cause https sessions to drop? even without packet loss? One theory is that packets may be arriving out of order and breaking the TCP connection in some way. We are using pingplotter to record data for us presently and we will be trying to match up failures with the flap events.
No comments:
Post a Comment