smbfs glitch

Urban Widmark urban at teststation.com
Sun Jan 21 19:37:27 GMT 2001


On Sat, 20 Jan 2001, Charles Howes wrote:

> On Sat, 20 Jan 2001, Urban Widmark wrote:
>
> > What kernel are you using? Do you get error messages from the kernel when
> > this happens? Does 2.2.18 work any better for you?
>
> The output is:
>
> # ls /ftp
> ls: /ftp: Input/output error
> session setup failed: code 0
> SMB connection failed

Hmm, wonder why it fails to connect ...

You could try one of these (depending on your samba version)
http://www.hojdpunkten.ac.se/054/samba/smbmount-2.0.7-ascii+fixes.patch
http://www.hojdpunkten.ac.se/054/samba/smbmount-2.2.0-alpha1-ascii+fixes.patch

They fix a bug in smbmount when reconnecting, if the reconnect fails
smbmount will SIGSEGV on a null pointer. They also send output to
log.smbmount (assuming %m.log) so you could mount with debug=4 (or
something) and see if it writes anything before it dies.

> And the log says:
> Nov 30 17:14:14 puma kernel: smb_trans2_request: result=-32, setting invalid
> Nov 30 17:14:19 puma kernel: smb_retry: signal failed, error=-3
> Nov 30 17:14:19 puma last message repeated 17 times
>
> The problem occurs about once every two months.  Can you suggest anything
> to put into the cron job to dump all relevant data?  Since it's a live
> system, I'm not willing to leave it half-mounted until I get around to
> fixing it.

You didn't answer my question, which kernel version? This should be better
in 2.2.18 (I am assuming samba 2.0.7 ...). It may fix this for you, it
does for some but other still see problems.

Someone has been using a cronjob to write to a file on the share once
every 10 minutes to stop it from needing to reconnect.

/Urban





More information about the samba mailing list