Re: [rtl] RT-Linux instrument control: Reboot/fsck question

[Randal Carpenter <wade@xwin.net>]

On Wed, 15 Apr 1998, Bill Blackwell wrote:

> I've been using RT-Linux for some time now for instrument control. 
> Everything works fine with one small exception, and it's not even RT-L
> related, but I thought someone here might be able to help.  It turns out
> that power outages (beyond my control) are causing file system problems.
> When fsck runs on the subsequent bootup, I have to interactively correct
> the problems.  However, the instrument is in a remote location, so I
> can't be there to guide fsck.  So the question is:
> 
> a)  Is there a good way to make the filesystem more robust with regard
> to spurious reboots?  A sync option when I mount / perhaps?
I tend to use a ups with a ups daemon running, cuts down on those reboots.

> 
> b)  Is there a way to get fsck to run non-interactively EVEN when it
> finds errors?

If I can get to a machine remote by whatever means, net dialin, etc and it
is at least finally coming up I do this for example

fsck /dev/hda1

and say yes to fsck mounted partition

fix it...
run sync 
then immediantly run

reboot -n -f


and when it comes up it normally finishs a good fsck on its own...

You could possibly automate this...when the error condition in the
/etc/rc.d/rc.S or whatever you use it started.  This could be done right
where fsck normally kicks off to an emergency login prompt.

Only I dont know how to force the Y to do mounted file system, piping a Y
in may work, if not make a custom fsck that dont ask that :)


--- [rtl] ---
For more information on Real-Time Linux see:
http://rtlinux.cs.nmt.edu/