Re: [gentoo-user] fsck won't work if ac cord not attached?!

[Willie Wong]

On Mon, Nov 30, 2009 at 06:54:45PM -0700, Penguin Lover Maxim Wexler squawked:
> you guys are killing me -- the problem goes away when the ac cord is
> plugged in. I open files watch videos surf the web and so on -- no
> problems. I'm no expert, but that would seem to suggest that the fs is
> OK, no? I set this forth above. Did your eyes glaze over at that
> point?

Nope. But please clarify if I remember wrong, since I have been only
half-following this thread since the beginning:

(a) When AC cord is plugged in, fsck runs on boot. 
(b) When running on battery, fsck refuses to run on boot. 
(c) When fsck does not run, your computer refuses to mount /var and
    /home?

Are all three of the above assertions correct? 

If not, please correct our impressions. If yes, then what Alan and
Neil said are perfectly reasonable: 

(i) You have a broken ext2 file system. Probably marked dirty from a 
  bad unmount prior to shutdown. 
(ii) On boot, when the AC cord is plugged in, fsck runs, so any error
  is fixed, and if no error, the file system is marked clean again.
(ii') When running on battery, because devs don't want fsck to run half
  way and have the computer run out of battery (which may corrupt the
  FS beyond whatever state it is already in), fsck does not run. 
(iii) Since the file system is marked clean, when the AC cord is in,
  the system boots fine. Directories are mounted, you can use it as
  usual. 
(iii') When the AC cord is out, the file system is still marked dirty,
  since fsck did not have a chance to look at it. Mount refuses to
  process those directories because Bad Things (tm) can happen. So
  your boot fails. 

Again, if (a-c) are correct, then what Neil and Alan said does NOT in
anyway contradict your observation I quoted just above; in fact, your
quote seems to make their diagnosis even more reasonable. 

According to what I vaguely remember of this thread (again correct me
if I am wrong), you see the symptom that (iii) behaves differently
from (iii'), and want to fix it by making its immediate causes (ii)
and (ii') agree. What Neil and Alan are telling you is that (ii) vs
(ii') should never be a problem (and I agree: on my Gigabyte netbook
my ext2 and my ext3 partitions never showed any behaviour like yours),
and in fact it is probably by design. That the reason why (iii) and
(iii') differ is actually (i). 

If you think this analysis is incorrect, please point out exactly
where my assumptions went awry. 

Cheers, 

W

-- 
"Unfamilliarity does bring terror, so I sympathize with those of you who 
aren't."
~DeathMech, S. Sondhi. P-town PHY 205
Sortir en Pantoufles: up 1089 days,  8:35