Howdy folks, I am running Knoppmyth Release 5 Alpha 12. Noticed this morning some odd behaviour, the storage was pretty much filled on the data/video drive. Mythweb wouldn't let me delete items and I couldn't watch live tv. I went to another machine and was able to delete items, but still no live tv. I noticed some other odd behaviour so decided to reboot and run a disk check. Well on reboot I got the following error

Loading LinuxEBDA is big; kernel setup stack overlaps LILO second stage.

That's a new one for me.

Googling pointed me to this link

http://nslug.ns.ca/pipermail/nslug/2004 ... 06479.html

I then booted off a Knoppix 3.9 cd and tried to repair it using chroot and the lilo -r but still got various errors, poking into the /boot directory on the filesystem shows the vmlinux-2.6.9-chw-4 to be of zero bytes. I don't see anything else in the /boot dir that is big enough to be the kernel, so either there is a broken link to the kernal image outside /boot, which I think is pretty unlikely or the kernal image is toast

So, 2 questions

1) Will fixing this be as simple as just copying the appropriate vmlinuz file off the R5A12 cd and replacing the zero byte one on the disk?

2) Would it be easier to just boot from a newer cd, say 5A16 and do an upgrade? Several times in the past I've tried to do an upgrade only to find it didn't work and my database of current shows wiped out. Most likely as my setup is non standard, the /myth and cache directories are on other hard drives etc.

Considering I have almost 150gb of shows, it would be a bad thing to have to start all over again with 150gb of unlabelled shows!

You would also need to rerun lilo so it can load the correct sector map for the replaced file. Unfortunately I don't know if the one that gets installed in /boot is identical to the one on the CD... Cecil should be able to answer that one.

Before you can do a good upgrade you need a good backup, especially of the database. Unless you've already got a current one you need to be able to boot the system to do this... I could do this myself, but just the thought of trying to talk someone else through it makes my head hurt.

Well, yes, fixing it was just that simple!

Booted off the R5A12 cd, started in expert mode, when it came to the
knoppmyth installer switched to another VT, mounted the root partition on
the hard drive, copied the same kernel image off the cd to the hard
drive's /boot directory, ran lilo -r /mnt/hda1 , reboot and life is good
again in the mythtv word.

Ahh sweet success, just in time to record battlerstar gallactica.

Hi sibblecomp,

Strongly suggest to do a backup fairly regular, and at least before a shutdown. Then if you decided to to do an upgrade or need to do an upgrade your database and content will be safe.

The backup of the database is saved on the /myth partion regardless of where the physical drive is located. ie, standard is hda1 for / and hda4 for /myth however hda1 / for root and hdd4 for myth works also. (depends where cd is located)

I have a small 16 gig with a standard install & upgrades (R5A16) and a much larger second drive for the media. After all is back up and running, I just edit the fstab to switch my second drive hdd4 to mount onto /myth and disable hda4. Then do a manual restore, reboot and it is like I never left. (It is hdd4 because it had R4V5 long ago)

Note, upgrade / install deletes any changes in the XF86Config-4 so you may want to manually save those to the /myth somewhere. Just to be safe, save a copy of any other changes in the /myth also. I made a directory /myth/tools just for that purpose.

Once you have the fstab setup to look at a second drive, you could in theory, use a removable tray to swap drives and build a very large archive. Just do a backup on power down, switch and then do a restore on power up to pick up the saved database. Sort of an offline "cold storage"

Also as a hint, if you have been running for a considerable time, there is probably some deadwood built up in the db and the /myth/tv. There is an enhanced script of mythlink.sh (pretty) that will help you clean up so you don't generate big error logs and bring the machine to it's knees. I think it is here in the General part of the forum.

Mike

I tried following this. I copied everything from /boot (symlinked to /UNIONFS/boot) to /media/hda1/boot. But then with lilo -r /media/hda1 (and /mnt/hda1 which is symlinked to /media/hda1) I get:


ideas?

_________________
LOL! "...amazingly, incredibly, spectacularly bad." -tjc

How about

I had tried that too, and got the same result.

I was just doing but doing got me past that issue. Unfortunately, I still don't get past

There are no error messages. It just stops.
I'm lost.

_________________
LOL! "...amazingly, incredibly, spectacularly bad." -tjc