On 9/9/05, Michael Segulja <[EMAIL PROTECTED]> wrote:
It only takes 5 seconds if nothing the the FS was actually corrupted. Otherwise, it will take a while.
Cool enough, someone else can answer. I don't let that happen to my file systems. As a workaround, you could code a tiny script that automatically jfs_fsck's your file systems at boot, if they are not mounted, and then mounts them.
Hrm... I have linux boxes with 280+ days up time. Granted, they are not running Myth, but usually apps don't crash the OS. Myth may be another thing since it is still young.
One thing that could replace a reset button if you wanted is a 'watchdog' ... these come in software and hardware. The hardware is more reliable but the software works pretty well most of the time too. Basically they run in the background and if they see the system become unresponsive, they reboot it.
You could also enable the "Magic SysRq Key" in your kernel. Then you can hit a powerful key combination to reboot even after a crash. Just figured I'd throw some options out there...
Someone else mentioned lockups as a fact of life for developers, but sometimes there is another way. Kernel developers, and people who test code that crashes a lot sometimes use VMware or User Mode Linux (kinda a poor man's VMware) which allows running one or more Linux OSs on top of the real/host Linux OS. These can be used for testing/modeling networks (emulate a whole network on one computer) but they are also very good for crash prone development and testing. At least you don't have to wait for the whole PC to reboot, go through BIOS checks, etc., just fire up another virtual instance after a crash. Don't know if this would work for Myth development, but figured I'd mention it.
Depends on the wife I guess :) ... my wife thinks its cool that we can watch TV uninterrupted when the power blinks out for a few minutes. Yes I have a UPS in the living room. My particular model isn't much uglier than the rest of the A/V gear.
I think you're missing the point I made in my email regarding kids &
reset buttons. The issue isn't the kids resetting the computer. The
issue is that when the system is not shut down cleanly, my JFS
filesystems won't mount automatically until I run jfs_fsck on them
manually. It only takes about 5 seconds to run and then mount them, but
it's a pain when it happens and I don't know about it because shows get
recorded to the wrong place.
It only takes 5 seconds if nothing the the FS was actually corrupted. Otherwise, it will take a while.
My 2 cents on JFS had to deal with this issue, not the fact that a
filesystem should be perfect no matter what happens to the PC. I'm
curious to see if anybody else has seen this behavior, because if not, I
probably just have a bad install or something.
Cool enough, someone else can answer. I don't let that happen to my file systems. As a workaround, you could code a tiny script that automatically jfs_fsck's your file systems at boot, if they are not mounted, and then mounts them.
> How about opening up the case and disconnecting the reset switch from
> the Mobo?
>
Disconnecting the reset button also means I can't reset the PC when I
need to in the odd times the whole system freezes, I do still have to
turn it on and off sometimes. :-)
Hrm... I have linux boxes with 280+ days up time. Granted, they are not running Myth, but usually apps don't crash the OS. Myth may be another thing since it is still young.
One thing that could replace a reset button if you wanted is a 'watchdog' ... these come in software and hardware. The hardware is more reliable but the software works pretty well most of the time too. Basically they run in the background and if they see the system become unresponsive, they reboot it.
You could also enable the "Magic SysRq Key" in your kernel. Then you can hit a powerful key combination to reboot even after a crash. Just figured I'd throw some options out there...
Someone else mentioned lockups as a fact of life for developers, but sometimes there is another way. Kernel developers, and people who test code that crashes a lot sometimes use VMware or User Mode Linux (kinda a poor man's VMware) which allows running one or more Linux OSs on top of the real/host Linux OS. These can be used for testing/modeling networks (emulate a whole network on one computer) but they are also very good for crash prone development and testing. At least you don't have to wait for the whole PC to reboot, go through BIOS checks, etc., just fire up another virtual instance after a crash. Don't know if this would work for Myth development, but figured I'd mention it.
Oh yeah, and a UPS connected to my HTPC that is in my living room
connected to my TV isn't the most attractive thing in the world. WAF
factor way down there....
Depends on the wife I guess :) ... my wife thinks its cool that we can watch TV uninterrupted when the power blinks out for a few minutes. Yes I have a UPS in the living room. My particular model isn't much uglier than the rest of the A/V gear.
I'm thinking of making a diskless front end via NFS, mounting all the file systems read-only. In that case, I will be able to reset until the cows come home, or just press the power button when I'm done watching TV and never worry about file system corruption.
Good luck with the JFS stuff.
_______________________________________________ mythtv-users mailing list mythtv-users@mythtv.org http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
