Frank Filz

> @ffilz I asked the libnfs maintainer, and he replied that it does not support 4.1 [sahlberg/libnfs#419](https://github.com/sahlberg/libnfs/issues/419) I guess it must be incomplete. but there must be some effort towards...

It was pointed out to me that PROXY does not use ntirpc for managing the send/receive. It would be good to move in the direction of having one send/receive and...

Something is probably corrupting memory with a use after free. A tcpdump might be interesting to show if there is some unusual request that might not be well tested. Otherwise,...

It would really help to have a tcpdump trace showing the stall. Also, seeing a difference between GigE and slower wireless, you aren't by chance using UDP?

Bump to keep on page 1

Hmm, I'll have a look at the tcpdumps. But if something is causing bad checksums, that would definitely cause problems (and make it look like the UDP problem - where...

Hmm, maybe the problem can also happen with TCP https://datatracker.ietf.org/doc/rfc4963/ Though historically we observed the issue with regularity on high speed networks with UDP, and switching to TCP resolved the...

Yea, almost all of the packets from 198.192.0.247 (bender, client?) have bad tcp checksum. On the other hand, Ganesha does seem to be responding, so I'm not sure they actually...

I don't see anything that suggests Ganesha is doing anything to cause a stall. I see a COMMIT, a NULL, and a SEQUENCE op all getting responded to. Maybe the...

Yea, that might help. If somehow you can get client and server in a way that's easy to match up between them, that would be very interesting.