rsync 2.4.6-3 cygwin : transfer interrupted (code 22) at util.c(211)

[Rönnblom Janåke /Teknous]

Having same problem as in previous threads with large number of files,
500.000+ and 20GB of data.

Im trying now to sync from w2k to a linux to see if I can get that working
before going to w2k -> w2k.

The client is running the rsync-cygwin-2.4.6-3 and the unix machine is 
running a trustix (a distrib) supplied package of rsync-2.4.6.

It works from another client with the same cygwin version but with a
smaller
number of files.

I get this on the client:

readlink home/Elev/00/jka1125/Favoriter/bhutan/ABC Country Book of Bhutan
- geog
raphy Flag, Map, Economy, Geography, Climate, Natural Resources, Current
Issues,
 International Agreements, Population, Social Statistics, Political
System.url:
File or path name too long

But it runs a for while after that before crashing with this message in
the log
on the linux machines:

transfer interrupted (code 22) at util.c(211)

a strace of the process just before the crash gives me this:

time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, "\377\377\377\377", 4)          = 4
time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, "\374\2004:>\34\5P", 8)         = 8
time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, ".BAK", 4)                      = 4
time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, "m", 1)                         = 1
time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, "\3", 1)                        = 1
time(NULL)                              = 1013533658
select(6, [5], NULL, NULL, {10800, 0})  = 1 (in [5], left {10800, 0})
read(5, "\0\0\262\313", 4)              = 4
time(NULL)                              = 1013533658
select(6, NULL, [5], NULL, {10800, 0})  = 1 (out [5], left {10800, 0})
write(5, "R\0\0\10overflow: flags=0x6d l1=3 l2"..., 86) = 86
time(NULL)                              = 1013533658
select(6, NULL, [5], NULL, {10800, 0})  = 1 (out [5], left {10800, 0})
write(5, "-\0\0\10ERROR: buffer overflow in re"..., 49) = 49
time(NULL)                              = 1013533658
rt_sigaction(SIGUSR1, {SIG_IGN}, {0x80501bc, [], SA_RESTART|0x4000000}, 8)
= 0
rt_sigaction(SIGUSR2, {SIG_IGN}, {0x80501d4, [], SA_RESTART|0x4000000}, 8)
= 0
getpid()                                = 29147
time(NULL)                              = 1013533658
write(3, "2002/02/12 18:07:38 [29147] tran"..., 74) = 74
munmap(0x2aabf000, 4096)                = 0
_exit(22)                               = ?

This seems to point to flist.c ...

Is there any connection between the inability on the client to open a file
(due to
long paths) and the crashes i experience? All my tests have these messages
about files that it can't open, even when using 2.5.3pre...

Next im going to see if I can get  a strace before it crashes on 2.5.3pre
and see
what that message is all about.

=====================================================
Janåke Rönnblom
SKERIA Utveckling AB (Teknous)
Assistentgatan 23
931 77 Skelleftea (Sweden)
-----------------------------------------------------
Phone  : +46-910-585424
Mobile : 070-3970743
Fax    : +46-910-585499
URL    : http://skeria.skelleftea.se
-----------------------------------------------------