[OpenAFS] Disk cache not updating

Derrick J Brashear shadow@dementia.org
Mon, 10 Jan 2005 21:58:04 -0500 (EST)


You shouldn't cc me on the replies.

>>> inaccessable, though they are from everywhere else. 'fs checkservers'
>>> reports the server down on this client.
>>
>> can you tcpdump -s 1500 -vv port 7001 on the client while this is
>> happening?
>
> Interesting -- bad udp checksum?

your kernel provides it, so...

> 16:11:42.551001 IP (tos 0x0, ttl  64, id 3082, offset 0, flags [none], 
proto 17, length: 60) CLIENT.afs3-callback > SERVER.afs3-fileserver: [bad 
udp cksum 4bc3!]  rx data cid 075159bc call# 23 seq 1 ser 66 
<client-init>,<last-pckt> fs call get-time (32)
> 16:11:42.551481 IP (tos 0x0, ttl  63, id 0, offset 0, flags [DF], proto 
17, length: 64) SERVER.afs3-fileserver > CLIENT.afs3-callback: [udp sum 
ok]  rx data cid 075159bc call# 23 seq 1 ser 44 <last-pckt> fs reply 
get-time 2005/01/10 16:11:42 (36)
> 16:11:42.551512 IP (tos 0x0, ttl  64, id 3083, offset 0, flags [none], 
proto 17, length: 93) CLIENT.afs3-callback > SERVER.afs3-fileserver: [bad 
udp cksum 157a!]  rx ack cid 075159bc call# 23 seq 0 ser 67 
<client-init>,<slow-start> first 2 serial 0 reason delay ifmtu 5692 maxmtu 1444 rwind 32 maxpackets 4 (65)

well, it's getting an answer which it seems to like, i have never seen 
this problem before. if it's reproducible i can suggest a patch to try, if 
you're willing.