In the last episode (Apr 01), pluknet said:> tcpdump'ed from RELENG_7, kernel and modules as of Mar 22 are in sync,
> world as of Mar 18.
>
> I caught this while building kernel via NFS. The subj host is an NFS
server.
>
> 23:22:03.056098 IP (tos 0x0, ttl 64, id 26932, offset 0, flags [DF], proto
TCP (6), length 160) 172.17.5.168.2355868444 > 172.17.5.167.2049: 108 getattr
[|nfs]
> 0x0000: 4500 00a0 6934 4000 4006 6db2 ac11 05a8
> 0x0010: ac11 05a7 0396 0801 7631 7eb4 37c6 db06
> 0x0020: 8018 40cc 54c6 0000 0101 080a 00a5 d95b
> 0x0030: da0f fa92 8000 0068 8c6b b31c 0000 0000
> 0x0040: 0000 0002 0001 86a3 0000 0003 0000 0001
> 0x0050: 0000
> 23:22:03.056180 IP (tos 0x0, ttl 128, id 10841, offset 0, flags [DF], proto
TCP (6), length 168) 172.17.5.167.2049 > 172.17.5.168.2355868444: reply ok
116 getattr [|nfs]
> 0x0000: 4500 00a8 2a59 4000 8006 6c85 ac11 05a7
> 0x0010: ac11 05a8 0801 0396 37c6 db06 7631 7f20
> 0x0020: 8018 71c7 7a07 0000 0101 080a da0f fa93
> 0x0030: 00a5 d95b 8000 0070 8c6b b31c 0000 0001
> 0x0040: 0000 0000 0000 0000 0000 0000 0000 0000
> 0x0050: 0000
>
> If you'd look in tcp header then 2355868444 value should be 918
actually
> (as htons(918) returns expected 9603 in hex (0396 in net order)).
That's not the port number; that's the NFS transaction id. See the
tcpdump
manpage, under the section "NFS Requests and Replies".
--
Dan Nelson
dnelson@allantgroup.com