RE: Network mounted file systems periodically lose connection
>> <Rob.Brenart@tradingtechnologies.com> wrote:
>> >> I have several network shares (windows) mounted as such
>> >>
>> >> //machine/share /mnt/mountpoint smbfs
>> >>
credentials=/credfile,rw,user,gid=sambawrites,fmask=0774,dmask=0774 0
>> 0
>> >>
>> >> And on occasion the share just dies. I'm guessing the reason for
the
>> >> death is something to do with our network, maybe a hiccup, or the
>> >> windows machine gets rebooted, or whatnot... in reality, I'm not
too
>> >> concerned with the why though. What I want is resiliency such that
>> the
>> >> links are immediately re-established. As it stands now, the mount
>> >> doesn't get unmounted or any such thing, I just get incredibly
awful
>> >> performance if I try to access the share, and eventually it will
>> timeout
>> >> and fail.
>>
>> I'm running sarge, off a net install.
>>
>> uname
>> Linux devcentral 2.6.8-1-386 #1 Thu Nov 25 04:24:08 UTC 2004 i686
>> GNU/Linux
>> >
>> dpkg
>> ii smbfs 3.0.10-1 mount and umount commands for the
smbfs
>> (for kernels >= than 2.2.x)
>>
>Well, I assume all is in order (sarge being newer then woody). My
>shares would usually disconnect after an hour of idleness, but it
>would reconnect if need be. Only a momentary delay while it's
>reattaching. (eg, server side connection idle timeout settings)
>
>How many shares and locations are you using? Are you resharing any of
>these? How much data I/O do you see on these, and what time frames?
>Do you have multiple linux machines with the same behavior?
> (experience tells us, we continue to make the same mistakes/oversights
>until we learn from them)
>
>Maybe some statistics polling used with MRTG could prove helpful if
>you're into that sort of thing. (carp-de exercise)
>
>If anyone else has ideas, feel free to interject.
I'm running 8 shares, they have different amounts of I/O...
For example, one sits idle most of the time then transfers over files
between 25-50megs in size for apache to redistribute.
3 of them sit idle 99% of the time, then send over a couple gig at 3AM
And so and and so forth... but the kicker is, I don't see any pattern
about which ones turn off when.
Reply to: