Gian Paolo Mureddu
gmureddu at prodigy.net.mx
Sun Aug 26 06:10:55 UTC 2007
Jon Nettleton escribió:
> On 8/25/07, Gian Paolo Mureddu <gmureddu at prodigy.net.mx> wrote:
>> Rui Tiago Cação Matos escribió:
>>> On Qua, 2007-08-15 at 15:56 -0400, Matthias Clasen wrote:
>>>> As discussed in the meeting, here is my personal
>>>> "laundry list" of low-hanging fruit. Note that some of
>>>> these are being worked on for F8 anyway.
>>> [snip sensible list of low-hanging fruit]
>>> I'd like to add another one:
>>> * get rid of the forced filesystem check every 26th mount.
>> Never heard of "tune2fs"? Just use it to set the mount count higher,
>> `man tune2fs`
> Okay, I am going to politely ask for an end to this sort of response
> in this mailing list. Especially in this thread we are brain-storming
> on ideas of a desktop fedora spin. Responses of RTFM or google
> it or whatever are not productive to the purpose of this list.
> I have heard of tune2fs, and probably every other obscure command
> sitting on our distribution's filesystem. We are brain storming ideas
> of what we do and don't want in an experimental spin of Fedora. This
> might sound harsh, however I want to put an end to it now. This is not
> Slashdot. If you have a reason that ever 26 reboots on a 300gig
> file-system a desktop user needs to wait 10 -15 minutes to wait
> while their journaled filesystem is forced checked then speak up.
> Otherwise head over to slashdot or digg of wherever and voice your
> I am finished now.
Simple enough take out the 1 (or 2's for additional filesystems) and
replace them with 0s in fstab ;)
LABEL=/ / ext3 defaults 1
0 -- from 1 to 0
LABEL=/boot /boot ext3 defaults 1 0 --
from 2 to 0
Simple, huh? That second check will take away the "mount count fs check"
but may also take away other features... From the man page....
"The sixth field, (fs_passno), is used by the fsck(8) program to
mine the order in which filesystem checks are done at reboot
root filesystem should be specified with a fs_passno of 1, and
filesystems should have a fs_passno of 2. Filesystems within a
will be checked sequentially, but filesystems on different
be checked at the same time to utilize parallelism available
hardware. If the sixth field is not present or zero, a value
is returned and fsck will assume that the filesystem does not
Default to 0 seems like what is being discussed, without tampering with
tune2fs... I don't like the approach (as that also takes away an extra
layer of protection for the hardware and data), but that *will* get rid
of the mount count fs checks. Alternatively, force Disk Druid to run
tune2fs with the -C 0 option to disable mount-count check on the
volume... That change, though I'm not sure would be as simple, though
might be the way to go in the long run.
More information about the desktop