[mythtv-users] catastrophic failure when disk fills up.

Misty P mistyp at thekorn.net
Thu May 22 15:57:55 UTC 2008


Chris Rouch wrote:
(...snippy...)
> These are repeated over and over (> 100 times per minute) and so in no
> time /var was also full.
>
> It would be nice if myth could spot "No space left on device" errors,
> and terminate the current recording (and refuse to do any more until
> there was space again). In this instance this would have fixed the
> problem without any intervention from me. Even if the autoexpire list
> was empty I'd only have to manually expire recordings to get things
> moving again, rather than having to manually clear space in /var.
>
> At the very least it would be better if error reporting in this case
> happened once rather than over and over.

I had something similar happen to me.  I can't remember the exact condition 
(it wasn't disk full), but the front end was logging massive amounts of 
information super quickly, and eventually ran /var out of file space.

I know there's a way to rate limit how many messages get logged, but can't 
remember how off the top of my head.  (it's on my to-do list)  Hopefully 
someone here knows how and will save me a few hours of google searching.  :)

Failing that, also on my to-do list is to move /var to its own partition, so 
that the next time myth decides to blow ten gigs of log data in five 
minutes, it'll just blow up /var and not / .

(That's a good idea on its own, FWIW.)

> There used to be an option to tell myth that so much of the disk was
> reserved for other use, but I couldn't find it - has this gone now?

That's still in there, but myth only checks once every 15 minutes IIRC.  So 
unless you're a *really* lucky guy, it probably wouldn't have saved you in 
this case.




More information about the mythtv-users mailing list