I had what appeared to be a frontend lock-up. The clock on the VFD was curent and blinking away; but the gui was dead and not going anywhere.
I got concerned when I couldn't ssh. I couldn't ctrl-alt-backspace. or anything. A hard reboot seemed to restart but left me with a Read-only / partition, and a basically useless mythbox.

For the first time I've seen, fsck ate my partition. There are literally hundreds of files in lost+found. I was able to find my mythconverg directory in lost+found, and the database files themselves appear to be intact, but as I indicated the rest of the file system seems to be a total loss. My last backup is a couple of weeks old (right after my 5.5 upgrade) so I'm hoping recovering from the raw files is OK.

So first question:
can I "upgrade" then overwrite the mythconverg directory with the backend down? or am I oversimplifying? maybe I can get in with the boot cd, and use it to run a mythbackup then get rid of the fs tarball?

Second: WTF? This was a clean 5.5 install on a new SATA disk on a new Gigabyte GA-M57SLI-S4 mobo. Maybe a defective disk?

At this point I have to shutdown the box before I do something (else) stupid. I've got my "sysadmin emergency kit." -- a cold 6 of Killian's.

I mentioned that this was a new system, the old system died from overheating after a northbridge fan went bad. So the whole mythbox concept has taken a major WAF hit recently. I need to score here just to preempt the "wouldn't a Comcast DVR be easier" argument. Well, the sysadmin emergency kit is starting to work, and I've already started rambling, so... Any advice?

This is a WAG (wild-ass guess) but I'm going to go with interrupts as the culprit. I think your disk system got confused with something else over which interrupt to use and your partition got roached in the process. When you look over the system, pay close attention to the interrupts used by the various devices and see if there are any shared interrupts which should not be.

Joe

_________________
Do you code to live, or live to code?
Search LinHES forum through Google

I had something very similar happen a month or two back, but with an earlier version of KM - I had some weird errors on my system, and fsck happily moved nearly everything into lost+found.

Boy was I unhappy when I learned that included the myth database.

I ended up recovering the database and various /etc files manually from lost+found before kludging a current backup and installing KM from scratch.

Note: This was soon after I had fiddled with the hdparm settings (not dma) for that drive; I blame that for my woes. The drive itself has been pretty solid ever since I put everything back the way it was...

As near as I can tell (by scanning through lspci -vv output) it appears SATA controllers are sharing interrupts with USB controllers. The contents of /proc/irq/*/ seemed to confirm.

First, what's the best way of detecting shared interrupts? my way seems a bit old school and convoluted...

Second, if SATA & USB are "shared interrupts which should not be" how do I go about reassigning? They're both "integrated peripherals" on the mobo, and I see no BIOS options for tweaking. --Last time I worried about shared IRQ's, you just had to move the jumpers around on one of the problem cards

Finally, I have the mobo listed above, imon vfd, a dvd burner, and 4 SATA drives, and a AMD X2 4400+ in the box. Is it possible I'm overloadnig a 650W power supply? (650W seems more than enough to me...)

Oh oh oh! Thank you! I was having similar system lockups on a new FE/BE system I just built and now that you mention power supply, I think you've just solved my problem. I started having the lockups after adding a 4th disk to the system.

*crosses his fingers*

Entirely possible. That's a pretty healthy pile of power eating hardware.

fudge. fudge-fudge-FUDGE!!!

OK, so how should I go about sizing a system power supply? I ran a previous incarnation of this system on a 450W. It only had 3 drives then, and a different CPU, RAM... still 650 seemed like a nice jump.

I just looked at the specs on my "spare" boot drive. It wants .72A on the +5V and .52A on the +12V. By my math that's 3.60W & 6.24W or just under 10 Watts. I'll assume they're all actually hungrier than this and call it 15Watts / drive. Further I'll assume not 4 SATA, and 1 PATA, but 6 SATA and 2 PATA (the mobo's total capacity.) So that's 8x15=120Watts.

If All 10 USB ports on the mobo were drawing the max of .5A@5V That's 5x10x.5=25Watts -- call it 30. If I assume 450W for the Mobo, RAM, Video card and CPU, that brings me to 600. Even assuming 25W apiece for my PVR-150s that's 650.

Now while I feel I'm being conservative, I realize just counting Watts doesn't account for the fact that the 5V and 12V buses are NOT interchangeable, so I figure my error is somewhere in there.

So how should I choose a Power Supply?

JourneySystems Power Supply Calculator -- looks like they've been redesigning it, but it should still work.

http://www.journeysystems.com/?power_supply_calculator

My issue ended up -not- being my power supply. It was a problem with the hald-addon-storage-polling locking up the system due to an incompatible old Compaq/HP DVDrom drive I put in the box. Swapped out the dvd drives and the problem vanished. Good luck with your issue. It does sound like power issues though.

They're redesigning it 'cause it's hoplessly out of date. It came up with just over 400 watts; but I'm sure they're short.

I've found something new.
I think the problem is unintentional overclocking. memtest is showing me errors in my RAM.
I think RAM errors can pretty much explain everything.

I think the discussion is moving into an area more appropriate for the Hardware forum, and less so here in "General Knoppmyth discussion" So I opened a new Topic HERE with all the details... see you there.

Thanks,

Joe