Izumi Tsutsui | 5 Nov 16:25 2008
Picon

5.0_BETA restorecd

ftp://ftp.NetBSD.org/pub/NetBSD/arch/cobalt/restore-cd/5.0_BETA/

See instructions.txt and CHANGES files
in tar.gz archive for details.
---
Izumi Tsutsui

Andy Ruhl | 5 Nov 16:28 2008
Picon

Re: 5.0_BETA restorecd

On Wed, Nov 5, 2008 at 8:25 AM, Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> wrote:
> ftp://ftp.NetBSD.org/pub/NetBSD/arch/cobalt/restore-cd/5.0_BETA/
>
> See instructions.txt and CHANGES files
> in tar.gz archive for details.

Thanks! I was just building a release but this makes the upgrade easier.

Andy

Izumi Tsutsui | 19 Nov 13:25 2008
Picon

updated Restore CD HOWTO page for 5.0_BETA

Hi,

I've updated NetBSD/cobalt Restore CD HOWTO page
for the newer 5.0_BETA restorecd script:

http://www.NetBSD.org/ports/cobalt/restorecd-howto.html

- update how-to-create section (using Makefile in the script)
- add a full "how to use" procedure
- add "Restore CD Internals" section for future developers

I've also upload updated the restorecd script and ISO image
based on netbsd-5/200811160002Z daily build snapshot:
ftp://ftp.NetBSD.org/pub/NetBSD/arch/cobalt/restore-cd/5.0_BETA/

Note now you can create your restorecd ISO image
even on Windows with Cygwin tools.

Enjoy,
---
Izumi Tsutsui

Andy Ruhl | 27 Nov 05:04 2008
Picon

release-5 restorecd issue

I'm not sure if this is fixed yet, sorry if it is.

I got the first version of the restore cd and did a complete overwrite
to a new drive on my Qube2 tonight.

When I tried to use chsh to change the shell, it complained that the
sticky bit wasn't set on /tmp.

I think this is common when unpacking the sets without using the
"preserve" option. If using tar, it's the -p option.

You can fix the problem by doing this:

chmod 1777 /tmp

Is this worth a pr?

Andy

Izumi Tsutsui | 27 Nov 14:29 2008
Picon

Re: release-5 restorecd issue

acruhl <at> gmail.com wrote:

> I got the first version of the restore cd and did a complete overwrite
> to a new drive on my Qube2 tonight.
> 
> When I tried to use chsh to change the shell, it complained that the
> sticky bit wasn't set on /tmp.
> 
> I think this is common when unpacking the sets without using the
> "preserve" option. If using tar, it's the -p option.
> 
> You can fix the problem by doing this:
> 
> chmod 1777 /tmp

Hmm, I wonder why no one else has noticed it for many years...

I'll prepare fixed one in this weekend,
but you could try the attached patch:

--- restorecd-5.0_BETA/data/cobalt/install/scripts/install-core.sh.orig	2008-11-01
17:34:39.000000000 +0900
+++ restorecd-5.0_BETA/data/cobalt/install/scripts/install-core.sh	2008-11-27
22:23:30.000000000 +0900
 <at>  <at>  -290,7 +290,7  <at>  <at> 

 	# Fix permissions
 	$CHMOD 755 /mnt/var
-	$CHMOD 4777 /mnt/tmp
+	$CHMOD 1777 /mnt/tmp
(Continue reading)

Hubert Feyrer | 27 Nov 14:50 2008
Picon

Re: release-5 restorecd issue

On Thu, 27 Nov 2008, Izumi Tsutsui wrote:
> restorecd isn't
> a part of official distribution (yet?),

I think it would be nice if we could extend the current job that build.sh 
does for a single job:

1) run "build.sh release" for all platforms
2) build things like the cobalt restore CD (which IIRC needs cobalt + i386
    binaries)
3) also build things like the multiboot CDs

Other things to add there may be to hook in some CDs with binary pkgs, 
live CDs, etc. but I think this needs some work on the general scheme of 
things first.

  - Hubert

Andy Ruhl | 29 Nov 03:51 2008
Picon

Qube2 with release-5 hangs, wapbl problem?

I have a Qube2 which I was running release-5 on before, but I recently
used the recovery CD to reinstall onto a new disk. The disk used to be
a regular 60 gig IDE, now it is a 30 gig laptop drive so I can unplug
the fan and use less power.

I wasn't having this problem before I used the restore CD. I noticed
that the default filesystem options include log now in the fstab,
which is why I susbect wapbl. I wasn't using the log option when I
installed release-5 manually before, because I never changed the fstab
or re-formatted the filesystem so I could use it. There have been no
errors logged to /var/log/messages about the new disk at all.

The system becomes generally unresponsive. I can't issue commands. My
VNC desktop is still visible over the network, I can type in a
terminal but as soon as I execute a command that terminal is hung. I
can't start a new one to issue more commands.

I have a serial console, but I can't log in with it once the system is hung.

I'm not hugely interested in debugging it at this point, but I can
probably do it in the next few weeks or so. What should I do? Maybe
set it to panic when I connect via the serial console or something?

My next step is to see if it's safe to remove the log option from the
fstab and reboot. Then I'll see if it hangs again. If not, I think
wapbl is probably a likely candidate. The hang isn't that easy to
generate, but it happens about once a day or so.

Thanks.

(Continue reading)

Izumi Tsutsui | 1 Dec 15:11 2008
Picon

Re: Qube2 with release-5 hangs, wapbl problem?

acruhl <at> gmail.com wrote:

> The system becomes generally unresponsive. I can't issue commands. My
> VNC desktop is still visible over the network, I can type in a
> terminal but as soon as I execute a command that terminal is hung. I
> can't start a new one to issue more commands.
> 
> I have a serial console, but I can't log in with it once the system is hung.
> 
> I'm not hugely interested in debugging it at this point, but I can
> probably do it in the next few weeks or so. What should I do? Maybe
> set it to panic when I connect via the serial console or something?

- try the latest 5.0_BETA kernel from ftp.NetBSD.org://pub/NetBSD-daily/
- try to send BREAK via serial console at freeze (to get ddb(4) console)
- try a custom kernel with options DIAGNOSTIC or DEBUG
etc?

> My next step is to see if it's safe to remove the log option from the
> fstab and reboot. Then I'll see if it hangs again.

As far as I tried on i386, it's safe if the file system
was properly unmounted on the previous shutdown.
---
Izumi Tsutsui

Andy Ruhl | 1 Dec 15:17 2008
Picon

Re: Qube2 with release-5 hangs, wapbl problem?

On Mon, Dec 1, 2008 at 7:11 AM, Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> wrote:
> As far as I tried on i386, it's safe if the file system
> was properly unmounted on the previous shutdown.

On Mon, Dec 1, 2008 at 7:11 AM, Izumi Tsutsui <tsutsui <at> ceres.dti.ne.jp> wrote:
> As far as I tried on i386, it's safe if the file system
> was properly unmounted on the previous shutdown.

So far, since I stopped using the "log" option in the fstab, the
machine has not hung for a few days now.

I have not been using the kernels from the NetBSD-daily stuff, I have
been doing a cvs update from netbsd-5 every day and when I need a
kernel, I build it from there. I did see the extra debug options in
the wapbl man page, I will add those and try again.

I rebooted the machine a few times when I was getting the hangs, so
the previous boot and mount of the filesystem should have been clean.

Thanks Izumi.

Andy


Gmane