[sshfs] reconnect issue?

[Hannes S. <hsc...@gm...>]

For quite some time now I'm desperately trying to set up a raid backup
server via sshfs, however after transferring an arbitrary amount of data
the backup script stops and the sshfs debug output looks like this (I
just included the interesting parts):

...
   [206839]         STATUS       28bytes (7688ms)
   [206840]         STATUS       28bytes (7688ms)
debug2: channel 0: window 216 sent adjust 4096
   [206841]         STATUS       28bytes (7688ms)
debug2: channel 0: rcvd eof
debug2: channel 0: o
   [206843]         STATUS       28bytes (7688ms)
   [206844]         STATUS       28bytes (7688ms)
...
debug3: channel 0: will not send data after close
...
debug1: Exit status 255
...
write: Broken pipe
    unique: 244261, error: -5 (Input/output error), outsize: 16
...

Looking at the log messages it seems the underlying ssh channel closes
because an eof is received, subsequently generating IO errors in the
sshfs binary. Any hints on that?

Usually the reconnect option works as expected when I manually try to
list or copy some files to/from the server after the failure. So the
question remains, why isn't sshfs reconnecting when receiving IO errors?

The error occurs with the following setup (gentoo ebuilds) and command:
kernel : gentoo-sources-2.6.15-r7
sshfs  : sshfs-fuse-1.6
fuse   : fuse-2.6.0-pre2
command: sshfs backup:/mnt/raid/ /mnt/raid -o
reconnect,allow_other,debug,sshfs_debug,loglevel=debug3,identityfile=~/.ssh/script_rsa

-- 
Hannes Schweizer
Digital RF Design
DICE GmbH & Co KG - Danube Integrated Circuit Engineering
Freistädterstr. 400, A-4040 Linz, Austria