Ask Your Question

Revision history [back]

loss recovery & sack issue

I don't know if the loss recovery phase seen on this capture is correct but it looks that retransmissions should be sent several times after receiving SACK and verify that the rest of packets haven't been received yet.

As you can check there are a lot of packet lost and SACK ( cannot handle all those gaps blocks ). on this case, retransmissions are only sent once and the partial ack and stretch sack handle the rest of the flow.

https://www.cloudshark.org/captures/1e22a0f6bfaf

Any idea how to solve this case? I know that BiF is so high but it typically works but in some situations like this.

Thanks in advanced.

loss recovery & sack issue

I don't know if the loss recovery phase seen on this capture is correct but it looks that retransmissions should be sent several times after receiving SACK and verify that the rest of packets haven't been received yet.

As you can check there are a lot of packet lost and SACK ( cannot handle all those gaps blocks ). on this case, retransmissions are only sent once and the partial ack and stretch sack handle the rest of the flow.

https://www.cloudshark.org/captures/1e22a0f6bfaf

Any idea how to solve this case? I know that BiF is so high but it typically works but in some situations like this.

in the tcptrace you can check how rare is the loss recovery phase. I don't understand it even i don't know if that is compliance.

Thanks in advanced.

loss recovery & sack issue

I don't know if the loss recovery phase seen on this capture is correct but it looks that retransmissions should be sent several times after receiving SACK and verify that the rest of packets haven't been received yet.

As you can check there are a lot of packet lost and SACK ( cannot handle all those gaps blocks ). on this case, retransmissions are only sent once and the partial ack and stretch sack handle the rest of the flow.

https://www.cloudshark.org/captures/1e22a0f6bfaf

Any idea how to solve this case? I know that BiF is so high but it typically works but in some situations like this.

in the tcptrace you can check how rare is the loss recovery phase. I don't understand it even i don't know if that is compliance.

Thanks in advanced.