rsync causing disk sleeps and loss of apps

[Richard]

> > 
> grep open rsync-debug.log | tail
> 3235  open("fromCDs/Documents/Frances/WorkTemplates/catterick newsletter", 
> O_RDONLY|O_NONBLOCK|O_LARGEFILE|O_DIRECTORY|O_CLOEXEC) = 3
> 3235  open("fromCDs/Documents/friendly.eps", O_RDONLY|O_LARGEFILE) = 3
> 3236  lstat64("fromCDs/Documents/Firewall/SurfNet_Firewall_files/hs-kader-
> openbaar.gif", 0xbfebe7bc) = -1 ENOENT (No such file or directory)
> 3237  open("fromCDs/Documents/friendly.eps", O_RDONLY|O_LARGEFILE) = -1 ENOENT 
> (No such file or directory)
> 3237  open("fromCDs/Documents/.friendly.eps.Np6jhQ", O_RDWR|O_CREAT|O_EXCL|
> O_LARGEFILE, 0600 <unfinished ...>
> 3237  <... open resumed> )              = 1
> 3236  lstat64("fromCDs/Documents/Frances/Thesis/20060613/opening of chapter 
> 3.doc", 0xbfebe7bc) = -1 ENOENT (No such file or directory)
> 3235  open("fromCDs/Documents/fstab101.txt", O_RDONLY|O_LARGEFILE) = 3
> 3235  open("fromCDs/Documents/ftp.jpg", O_RDONLY|O_LARGEFILE) = 3
> 3235  open("fromCDs/Documents/garden_top.jpg", O_RDONLY|O_LARGEFILE) = 3

nothing obvious here. Maybe there is something interesting between the last open() 
line and where it gets into the never ending select()/gettimeofday() loop:


> 3236  <... select resumed> )            = ? ERESTARTNOHAND (To be restarted)
> 3235  <... select resumed> )            = ? ERESTARTNOHAND (To be restarted)
> 3236  --- SIGINT (Interrupt) @ 0 (0) ---
> 3235  --- SIGINT (Interrupt) @ 0 (0) ---
> 3236  gettimeofday( <unfinished ...>
> 3235  gettimeofday( <unfinished ...>
> 3236  <... gettimeofday resumed> {1296658927, 336394}, NULL) = 0
> 3235  <... gettimeofday resumed> {1296658927, 336386}, NULL) = 0
> 3236  select(0, NULL, NULL, NULL, {0, 400000} <unfinished ...>
> 3235  select(0, NULL, NULL, NULL, {0, 400000} <unfinished ...>
> 3236  <... select resumed> )            = 0 (Timeout)
> 3235  <... select resumed> )            = 0 (Timeout)
> 3235  gettimeofday({1296658927, 737109}, NULL) = 0
> 3236  gettimeofday( <unfinished ...>

If it is long it should be very well compressible with bzip2 so you can post it 
here.  Did you look at /var/log/messages?

Finally, if eg dolphin is hanging you can find out its PID and attach strace to it
with "strace -p PID" and if you are lucky you will see on which syscall it is hanging..
although if it is hanging on dbus or something like that as I suspect it would not be
kentirely easy to recognise what exactly is happening. Also the output will be flooded
with X11 related syscalls.

> The comments about read-only made no sense to me, but it seems that they are 
> being written as owned nobody:nobody.  I guess that that would mean that at 
> the next write they can't be overwritten by files owned by anne:users - 
> correct?  Certainly there are lots of messages like
> 
> rsync: chgrp "/mnt/server_home/anne/Documents/Bills/fluendo.pdf" failed: 
> Operation not permitted (1)
> 
> and I couldn't see why it was wanting to change groups.

not sure, some filesystems do not support it, might be mounted with wrong options
but it certainly should not hang the system.


Richard

---
Name and OpenPGP keys available from pgp key servers