a-lunev commented on PR #6562: URL: https://github.com/apache/incubator-nuttx/pull/6562#issuecomment-1180124212
> @a-lunev > > > Have you tried testing #5589 namely over [1ded8bb](https://github.com/apache/incubator-nuttx/commit/1ded8bbabbe0b23e3242be17fb87d418966778b2) (Feb 22)? > > Yes, I also tried the code base you suggested, no one has changed the logic of fast retransmission during this time, so the behavior is consistent with the current master branch. Thank you for checking it. Some time later I could investigate the issue why the Fast Retransmit algorithm is not triggered based on your test setup. > > I am not just understanding why #6451 was done implementing a half logic vs #5589 implementing a full logic (according to RFC 6298 (5.4))? > > PR #5589 removed the logic of retransmitting all unack segments, which resulted in slower retransmission recovery, each RTO will only compensate for a segment loss, so you can see that the connection recovery took about 4.5s(3 * 1.5 RTT), but in In the current logic, if the retransmission segment does not get the expected ack after the fast retransmission is triggered, the RTT timeout will retransmit all the packets, which ensures the fast recovery of the connection. This is why I don't want you to change the RTT retransmission logic to only one the earliest not acknowledged segment. If this not standard algorithm is really important to you, possibly it had better to isolate it by a conditional compilation and menu option? I am sure #5589 should be normally used as it fully follows RFC 6298 (5.4). > ... you can see that the connection recovery took about 4.5s(3 * 1.5 RTT) ... The low recovery time you are referring is not directly related to #5589 (or #6451, #6562). The cause of the issue is that the Fast Retransmit algorithm was not triggered in your test. This is another task like #5598. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@nuttx.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
