response on the bottom.

regards,
Andre | http://www.varon.ca

On Fri, Sep 24, 2010 at 8:39 AM, Red Sancho <[email protected]> wrote:
> that's a long shot up sell but worth trying, we've already identified that
> risk of not having a backup policy but from where we stand right now adding
> cost to a losing business isnt gonna go far up the ladder... anyways, is it
> possible to prevent or at least detect an FS failure looming? This is what

As long as you have jounalling in the FS, Forget about the FS
monitoring. I have
files at home a decade ago stored in diverse filesystems like ext2, ext3, NTFS,
in different stages over the years with power outages in between.  Please note,
my point is for home users (not, power users) and to a certain extent,
SMB with tight
money problems. But for others, definitely backups or at least, mozy.

What you should check is the harddisk integrity instead via SMART attributes.
There are many software for Linux, but I’m partial to Harddisk Sentinel which
Gives you a high level percentage of your HDD fitness. Free for
Linux, nagware for windows. These will be your canary to the coal mines, a
Warning that your HDD is about to fail.

No offense for HW RAID users, that is why I use JBOD at home and not HW raid
for this so I can monitor it on the OS level.

> happend... after joining the company i found out that most of the servers
> are running on debian etch and i brought this to the attention of my boss
> and he agreed to perform upgrades to the majority of the etch servers. after

I resent that as a happy debian etch user even now! :) Will be using it for a
specific app for a very long time.  Joking aside, upgrades are a good thing
anyway but preparation is the key.

> performing the distro upgrade and rebooting, our sdb1 didn't mount and all
> of the files went into lost+found, i also got "missing journal for
> /dev/sdb1" from messages. i tried to mount it only to find that all of the
> files are now in lost+found, so i unmounted it and ran fsck (probably a bad
> thing). Now im wondering what could have caused this error in the fs.

Since you did not post logs data and only generic information from you, just a
Stab in the dark of your problem could be that:

  * fsck messed it up. I doubt it though, but could contribute to the issue.
  * When doing upgrades, you run apt-get updates or upgrade. Well, guess what…
your sources.list file is pointing “stable” to Lenny, not etch which was
was done by debian few years ago. This will present a HUGE upgrade.
and potential
issues from my experience.

A third likely scenario is you did a combo on the above. Upgrades from
etch to lenny
Probably contributed small issues, they tried to fix it by fsck. The
two small issues
That could be fixed individually, turned to a bigger issue.

The only time i have personally seen a smooth upgrade from one release
to another is
when doing weekly or month apt-get before a major release is going to happen.

If that is the case, then no way but to check each file and put it
back from lost+found. :(
or backups as others have mentioned.

FYI, what I do to upgrade is:
1. read the upgrades from one version to another on debian’s website.

2. run below. you might have to run it multiple times to satisfy dependencies.

  # apt-get update ; apt-get upgrade

Install latest debian etch updates, reboot next just to make sure.
Make sure you rename
stable to etch to explicitly point to etch. Note that etch is now
archived,  and only
available on archive.debian.org, so change the sources.list appropriately.

3. reboot  just to be sure it comes up. If it does not, you have
bigger problems that just
doing distro upgrades!

4. change sources.list to point to lenny explicitly, or just use stable.

5. then do upgrades:

  # apt-get dist-upgrade

You might have to run this multiple times to satisfy dependencies.

6. reboot and enjoy the latest debian new release.

Please note that it's a bit winding work, but been doing this for a
few years with multiple
Debian release and the steps above has served me well.

> also one question, is it ok to run fsck read-only on a mounted device?
No…
_________________________________________________
Philippine Linux Users' Group (PLUG) Mailing List
http://lists.linux.org.ph/mailman/listinfo/plug
Searchable Archives: http://archives.free.net.ph

Reply via email to