I understand the basics of file layout, inodes and inode/file-name mapping and such. But I still basically always hit yes to fix any FS errors! Is there any guide to educate on when to say no ;)
About the RFE. That sounds interesting. Having sshd into initrd would be useful. But what would be more useful is having the "actual" console messages/errors printed there. i.e. Is there a way to have that telnetd in busybox (or sshd) actually be a "virtual" console that shows errors, and accepts logins and system repair operations ? Let me know if that can be done somehow. Basically, it should be just like a serial console, just over sshd
umm .. maybe screen inside sshd ?
On Fri, Apr 18, 2008 at 12:07 AM, John Summerfield <
debian@herakles.homelinux.org> wrote:
Tom Sightler wrote:
On Tue, 2008-04-15 at 23:40 +0200, Ahmed Kamal wrote:
Problem
=====
This leaves me with no "remote" way to fix serious errors such as FS
checking
Suggestion
=======
Is there such a thing as remote serial console. I'm thinking *if* the
kernel gets to boot, then it immediately starts the network interface,
and a tiny server for sending console messages to whoever is
connected. I should be able to fix FS corruption and other serious
problems. My question is, does such a thing exist ?
If most of your problems are simply moderately serious filesystem
corruption that you repair by running fsck and answering "y" to all the
questions (which is what most people seem to do) then you can probably
just add the file /etc/sysconfig/autofsck with contents like:
AUTOFSCK_OPT="-y"
I've often wondered how many people are competent (as in able to make informed decisions) about the questions a manual fsck presents.
I still feel much the same as I did when I was presented with my first one, around RHL (not RHEL) 4.x or earlier:
??
<shrug>
e2fsck -y ...
xx fingers.
Seems okay.