[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Sarge Upgrade Broke Everything Except Terminal

Leonard Chatagnier wrote:


From: "Kent West" <westk@acu.edu>

Leonard Chatagnier wrote:

[System brokenness on upgrade]


I'm just coming into this, but it sounds like you've tried upgrading
both your system to Sarge, and your system to a new kernel. This warning
is a little late, but next time you might want to take these two steps
separately.

I don't understand this two step thing.  Please explain, maybe in a separate
reply so not to clutter this posting any more than necessary.
It's good to want to avoid clutter, but to make the information to other seekers, it's probably best to keep this on-list.
As a general rule, performing an "apt-get upgrade" or "apt-get dist-upgrade" (or equivalent) does not upgrade the kernel, unless a special kernel package has been installed which depends on the newest kernel (if I understand correctly). In order to upgrade your kernel, you'd have to do some additional step, such as "apt-get install <some_kernel_image>". 


I'm not sure what you mean by "virtual kernel-image". When I do an
"apt-cache search kernel-image-2.6.8", with my sources.list pointing at
unstable, I see no version -10; -9 is the highest I've seen. This
indicates to me that you're getting your kernel image from some
non-Debian source; in such a situation, breakage is not surprising to
me. (Of course, I might be misunderstanding something here.)

I'll explain. It's all debian, but not via apt-get as I've used it until no
more upgrades would list. The virtual package was found searching/browsing
the Debian site.  The URL and package name are:
http://packages.debian.org/testing/base/kernel-image-2.6.8-1-686
Package: kernel-image-2.6.8-1-686 (2.6.8-10)
Linux kernel image for version 2.6.8 on PPro/Celeron/PII/PIII/PIV

Okay, so the numbering difference was throwing me off. It's all good.


I don't believe that you should have needed to run "mkinitrd" unless
you've compiled your own kernel; at least I've never had to use this
utility when I install a standard kernel image.

I didn't compile. I've read plenty manuals since last post and saw no
reference to it. I came across it in my searches and it seemed aprapopo so I used it
with the -r option.  It never completed without error. But, when I rebooted after using
it, the new kernel loaded for the first time after many failed attempts.
"It seemed aprapopo so I used it with the -r option." You used what with the "-r" option? dpkg? Something else? 


What do you get when you run "uname -a"? It sounds like you've got a
broken kernel, and a broken/unfinished apt-upgrade. I'd try fixing the
kernel first, then worry about fixing the upgrade. Can you boot into
your old kernel?

uname -r gives Linux debian 2.6.8-1-686 #1 etc. It does appear to me that
some modules are loaded and unloaded.  There are some messages that appear
on the bootup screen that scroll by to fast to read and dmesg doesn't show
them. I wish I knew what log file contained them so I could see what the problem
it, but using locate to veiw log files hasn't picked it up yet.
Kernel messages are logged and are viewable by dmesg. Other boot messages, such as when the init scripts start running, may or may not be logged (probably not, by default). As long as you never leave the terminal on which the messages appear (such as by having a graphical login manual attempt to start, or by pressing Alt-F3, etc), you should be able to scroll back up through the messages with Shift-PgUp. Otherwise, you can turn on logging for the boot scripts, I believe, by editing "/etc/default/bootlogd" and setting it to "yes". I've never done this, so don't know the repercussions. 


Everything works
fine on the old
kernel boot.
In that case, I'd boot into the old kernel, and run "apt-get update" followed by "apt-get install kernel-image-2.6.8-1-686". This should successfully install your new kernel. (This assumes that your "/etc/apt/sources.list" file is pointing to a source that contains this package.) 


If your modem is a real modem, you can probably recover; if it's a
win-modem, that requires some driver, you may be looking at a world of
hurt, at least until you've got a properly working kernel. If your modem
is real, you should be able to set up a connection either with pppconfig
or with wvdial; once you have a working connection, you can finish your
upgrade.

I have a win modem Diamond MM SupraMax 56i Voice PCI modem and the drivers
are installed and the modem worked under Woody-bf2.4 and still does.  But be
forewarned, I will not use apt-get again until I get some help on why it hasn't already
worked for me after many updates, upgrages and dist-upgrades using the "testing" configuration
in my sources.list.
I'm getting tired of spinning my wheels gettin nowhere using apt-get and
I've read the manual(s).
I'll be happy to send you my source.list for examination.  I believe you
were the one who helped me in previous postings (Mewbie Having Much Trouble Getting Deb 2.4.18-bf2.4
To Work) on what's required after the mirror listing.

FLASH: New info.  I've discovered that my /var partition is full again
Ah! That'll definitely cause problems. I would recommend running "apt-get autoclean". If that doesn't free up enough space, try "apt-get clean"; this will erase any packages you've downloaded but have not yet installed, but it'll give you room to get your kernel properly installed, followed by a judicial upgrading that's easier on your disk space issues. 


and
probably prevented some
required files from downloading when I ran dpkg -i on the kernel-image.
Ah! So you did run "dpkg" to install your kernel image. "dpkg" is the bare-bones engine for installing stuff. Unless you know what you're doing, you're likely to be shooting yourself in the foot by using it. It would be MUCH better to use "apt-get install <kernel_image>". 


I've made a symbolic link for the apt-get cache in past to work around this.  The install must have filled
it up again so I need to create some more links to get a work around.  I wish I could repartition and
correct this problem but I don't want to have to reinstall everything again starting with Woody.  I have plenty
room on the hard drive. Is this the root of the problem?
I've found that a /var partition of 500MB to 750MB usually works fine for me, as long as I clean out my apt-cache every so often (using "apt-get [auto]clean" as mentioned above). This also assumes you're not running a bunch of var-filling services, like an email server for a bunch of folks with large attachments, etc.
If you've got a large empty partition somewhere, or a partition you can temporarily live without (like /tmp) which is big enough for another partition's files, you can move, unmount, mount, copy, delete, rearrange partitions, as long as you're careful. I've done it several times on different boxes, but it's easy to get confused about where your files are now and where they're going and when to mount/unmount/delete/copy, etc. Ideally you'd back up your system, fix your partitions, then restore, but I like to live more dangerously than that on my workstations. No risk, no thrill, and all that.
Reply to: