My timeouts when emacs moves (copies) files - my email to AFS

[Harald Barth]

Moves to AFS are interrupted with "network is down". 

Log: What's error 3791? And then the rx call times out :-(
I have no other ways to reproduce than to run mew in
emacs on my email :-( It does not happen all the time. I
suspect that I need to idle about the time it takes to
write one or two emails.


Mar 25 00:08:44 habarber xn_lookup: found node cc0c3400
Mar 25 00:08:44 habarber cache_name '00/2F'
Mar 25 00:08:44 habarber nnpfs_message_installdata dentry: ffffffea
Mar 25 00:08:44 habarber nnpfs_message_installdata: tokens: 0xd5
Mar 25 00:08:44 habarber nnpfs_message_installdata size before: 0
Mar 25 00:08:44 habarber nnpfs_message_installdata size after: 0
Mar 25 00:08:44 habarber nnpfs_message_receive opcode = 1
Mar 25 00:08:44 habarber nnpfs_message_wakeup error: 0 seq = 247
Mar 25 00:08:44 habarber nnpfs_message_rpc wakeup: seq = 434
Mar 25 00:08:44 habarber nnpfs_message_rpc opcode this_process->error_or_size = 
0
Mar 25 00:08:44 habarber nnpfs_message_rpc opcode ((nnpfs_message_wakeup*)(this_
process->message))->error = 0
Mar 25 00:08:44 habarber nnpfs_open_valid: exit tokens 0xd5
Mar 25 00:08:44 habarber nnpfs_write_file: tokens: 0xd5
Mar 25 00:08:44 habarber nnpfs_data_valid: offset: want 0 has 0, tokens: want 80
 has d5 length: 0
Mar 25 00:08:44 habarber nnpfs_write_file: error = 3791
Mar 25 00:08:44 habarber nnpfs_flush
Mar 25 00:08:44 habarber nnpfs_release_file
Mar 25 00:08:44 habarber nnpfs_release_file inode->i_count: 1 inode: d17cb368 al
iases: 2879(c1ff353c)
Mar 25 00:08:44 habarber find_pag: ngroups = 1
Mar 25 00:08:44 habarber find_pag: Did not find pag
Mar 25 00:08:44 habarber nnpfs_get_pag_group: returning 22421
Mar 25 00:08:44 habarber nnpfs_message_rpc opcode = 13
Mar 25 00:08:44 habarber nnpfs_message_rpc seq = 248
Mar 24 23:08:44 habarber arla[10700]: worker 0: processing
Mar 24 23:08:44 habarber arla[10700]: Rec message: opcode = 13 (put_data), size 
= 96
Mar 24 23:08:44 habarber arla[10700]: putdata (154.536875634.309564.261884)
Mar 24 23:08:44 habarber arla[10700]: copyfd2rx: rx - call timeout
Mar 24 23:08:44 habarber arla[10700]: writing back file: Network is down
Mar 24 23:08:44 habarber arla[10700]: nnpfs_message_putdata: cm_close: Network i
s down

The syscalls around the  same time are

access("/home/haba/Mail/folders/inbox/133604", W_OK) = 0
rename("/home/haba/Mail/folders/inbox/133604", "/home/haba/Mail/folders/sunet-fa
il/2879") = -1 EXDEV (Invalid cross-device link)
stat64("/home/haba/Mail/folders/sunet-fail/2879", 0xbfc7faa0) = -1 ENOENT (No su
ch file or directory)
open("/home/haba/Mail/folders/inbox/133604", O_RDONLY|O_LARGEFILE) = 5
fstat64(5, {st_mode=S_IFREG|0644, st_size=3791, ...}) = 0
open("/home/haba/Mail/folders/sunet-fail/2879", O_WRONLY|O_CREAT|O_TRUNC|O_LARGE
FILE, 0666) = 6
--- SIGALRM (Alarm clock) @ 0 (0) ---
gettimeofday({1143241724, 446838}, NULL) = 0
--- SIGIO (I/O possible) @ 0 (0) ---
write(4, "\2\0\4\0\36\0\240\0\0@\0\0(\0\240\0", 16) = 16
ioctl(4, FIONREAD, [32])                = 0
read(4, "\0035LY\305c\177\1;\0\0\0\35\0\240\0\36\0\240\0A\3\0\3"..., 32) = 32
ioctl(4, FIONREAD, [0])                 = 0
sigreturn()                             = ? (mask now [ALRM])
gettimeofday({1143241724, 447456}, NULL) = 0
gettimeofday({1143241724, 447504}, NULL) = 0
setitimer(ITIMER_REAL, {it_interval={0, 0}, it_value={0, 99334}}, NULL) = 0
sigreturn()                             = ? (mask now [])
read(5, "Return-Path: owner-sunet-fail at SU"..., 16384) = 3791
write(6, "Return-Path: owner-sunet-fail at SU"..., 3791) = 3791
read(5, "", 16384)                      = 0
close(6)                                = -1 ENETDOWN (Network is down)

Arla is curent and OpenAFS is 1.4.1rc10.

Harald.

PS: And my kernel has another timezone than arlad?