On Wed, Mar 06, 2019 at 05:59:39PM +0800, Jiufei Xue wrote: > Hi Trond, > > > > > No, that's quite weird... Were you able to catch this behaviour on > > wireshark? It would be interesting to see if the server is actually > > sending 2 SEQUENCE ops or if that is a corruption that occurs on the > > client. > > > > Many packages are dropped by kernel when I use tcpdump, so I am not sure > whether the server is sending 2 SEQUENCE. For what it's worth, tcpdump with -s200 or so should be enough to get this. Or maybe capturing from a different machine could help? > However, according to tracepoint nfsd_compound_status, the server is > processing SEQUENCE op only once. I think it is more likely blames to > the client. If there was a bug in nfsd's xdr decoding, that'd happen after that tracepoint. That'd be weird corruption, but perhaps all the possibilites are weird at this point. --b. > Is there any other way to catch the behaviour? > > Thanks, > Jiufei >
