xfs
[Top] [All Lists]

Re: 1.0 Installer

To: Eric Sandeen <sandeen@xxxxxxx>
Subject: Re: 1.0 Installer
From: Seth Mos <knuffie@xxxxxxxxx>
Date: Thu, 24 May 2001 21:43:17 +0200
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <3B0D5DEE.54EC5532@xxxxxxx>
References: <4.3.2.7.2.20010524204643.039126c0@xxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
At 14:15 24-5-2001 -0500, Eric Sandeen wrote:
Seth Mos wrote:
>
> The installer seems to barf when trying to Upgrade a installation of a
> system where the XFS filesystem is not clean.
> I had to boot the machine another time to recover the filesystem to a clean
> state before I could try to upgrade it.

Hm, we didn't anticipate this scenario.  :)  FWIW, the same thing
happens under ext2, if it detects that a filesystem was not cleanly
unmounted, the install won't continue.

Uhm, that shouldn't happen with a journaling fs :D

> It first says that it will to upgrade the linux system on hda6 Anaconda
> then throws me a error that there is no linux system on hda6 (which is the
> root fs).
>
> Although it may be hardware issues, the box is acting funny every now and
> then. It should not produce this error, but recover the fs on mount during
> install, would it?

I'm not sure if mount returns differently based on whether or not it
went through recovery... if so, anaconda may be interpreting this as a
problem.  I'd have to look into it... but for now, just run an upgrade
only on clean systems, I guess.

I will, it's one for the archives.

You say it's acting funny, do you mean _after_ the install?

Uhm no, before, during and after. Is that the right answer ;)
Seriously, I have seen the installer die with signal 11 when the cd drive was slave on the 2nd channel.
fsck.ext2 produced a signal 11 during boot when the /boot was clean.
It produced a signal 11 and a 4 during some other attempts after starting anaconda.

I will run memtest when it's upgraded.
The machine was a redhat 7.0 system that was installed at that time with the 2.4.0-test5 installer. And that one went fine :-)

The kernel that I had compiled on the machine does not misbehave that often.

It's weird though, I tried to make the machine fall over by compiling the kernel a 1000 times with make -j20 but it completed succesfully.
So for some reason the tide here seems to affect the stability of the machine.

> Just checking if anyone else has seen this behaviour.
> Does anyone know if this also happens when the ext2 fs is dirty or not?

Yep, the installer will ask you to reboot & fix the problem.

It did not ask me to reboot and fix the problem, it just reported there was no linux installation on /dev/hda6 and decided that enough was enough.

Cheers

--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.


<Prev in Thread] Current Thread [Next in Thread>