client: 10.x.x.17 server: 10.x.x.9
Average upload: ~ 600Mb/s (on 1gig link) Average download: only 40Mb/s QoS is not enabled.
I've been working with the vendor and pulling almost all my hair out for several months now but can't find the root cause.
Capture screenshot: [ https://a.tmp.ninja/eouvJopb.png ]
Any tips would brighten up my day!
You have packet loss at your capture point from .17 to .9. Frame 258, .17 to .9 the sequence number is 38 and the next expected sequence number is 1398. Frame 261, .17 to .9 the sequence number is 13638, not 1398. There is a big gap of missing frames. Frames 262, 264, 266, 268, .9 to .17. The .9 device, is requesting a packet with the sequence number 1398. Frame 271, .17 to .9 the sequence number is 1398, and the next expected sequence number is 2758.
The first step is to fix the upload problem is to fix the missing packets' problem. The upload from .17 to .9 slows until the sender sequence numbers and receiver acknowledgement sequence numbers are aligned. My suggestion is to move your capture point closer to .17, to isolate if the problem is the transport or TCP device.
Hi Big,
Thanks for the analysis.
From your point of view, who is dropping the packet?. 9 or. 17? Our vendor point out that the OS (. 9) is causing the drops but we have lots of laptops, we've tested with different operating systems, all laptops (. 9) showing same issue.
Also, what could have cause these missing frames?
Asked: 2022-03-15 14:39:37 +0000
Seen: 237 times
Last updated: Mar 15 '22
Copyright Wireshark Foundation, 2017-2023 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
