|
Grex > Jelly > #72: Stupid unix (solaris) question. | |
|
| Author |
Message |
mickeyd
|
|
Stupid unix (solaris) question.
| 
Mar 4 00:31 UTC 2007 |
So, I know this is a lame question. But, say for example you have a sun ultra
10, with two hard drives (ide). is there any point or need to run fsck on them
with any regularity, or only if they complain about needing it to be ran?
Just curious. thanks you.
|
| 10 responses total. |
mcnally
|
|
response 1 of 10:
|
Mar 4 01:05 UTC 2007 |
It's usual for Unix machines to go months or years without a reboot and
filesystems are usually only fsck'ed on boot, and even then usually only
get the full scan if they were unmounted "dirty." I wouldn't think you
would need to schedule fscks for your disks.
|
maus
|
|
response 2 of 10:
|
Mar 5 06:24 UTC 2007 |
Unless you suspect the filesystem has been damaged, don't bother.
Typical causes of damage are the machine shutting down without umounting
and syncing the filesystems (usually recoverable) or a pathological
programme scribbling directly on the slice/drive instead of by way of
the filesystem (usually not as recoverable). Note, you should only
manually fsck a filesystem that is not mounted. If the filesystem is
dirty or has not been fscked in n reboots, it will be automagically
fscked. otherwise, don't bother.
|
cross
|
|
response 3 of 10:
|
Mar 5 13:59 UTC 2007 |
I believe the every n reboots thing is a Linux-ism. Regardless, the
recommendation back in the day used to be to unmount the filesystem and fsck
it after large data movement operations; e.g., recovery from tape, system
installation, etc. These days, the filesystem has been tuned to an extent
that this is no longer necessary.
|
maus
|
|
response 4 of 10:
|
Mar 5 17:34 UTC 2007 |
The man-page for vfstab file implies that a filesystem can be considered
for automatic fsck (the fsck pass column in the vfstab file). A copy of
the man-page is at
http://www.cs.bgu.ac.il/~arik/usail/man/solaris/vfstab.4.html .
|
maus
|
|
response 5 of 10:
|
Mar 5 17:38 UTC 2007 |
http://www.joho.com/sun/ch04/131-133.html seems to confirm it.
On a particularly old version (such as Solaris 1.x (BSD-based)), I would
agree with Dan, it's a good idea to tidy up after a big data move, but
otherwise, the fs seems to be pretty much self managing. The new ZFS is
even nicer about such things (and just generally spiffy in so many
ways).
|
cross
|
|
response 6 of 10:
|
Mar 6 03:34 UTC 2007 |
Oh, does Solaris do it too? Hmm. Interesting. Or are you talking about the
fsck pass field? That's rather different; lots of OS's do a boot-time preen
fsck, but that's different from a full-blown fsck.
|
maus
|
|
response 7 of 10:
|
Mar 6 04:25 UTC 2007 |
I was thinking that the boot-time fsck was what he was referring to. I
know that if it fails, it just dumps (rather unceremoniously) into sh so
you can finish yourself the fsck.
|
cross
|
|
response 8 of 10:
|
Mar 6 13:37 UTC 2007 |
Right. That's the the pruning fsck; the totally automated one. It's rather
different than a `full' fsck.
|
maus
|
|
response 9 of 10:
|
Mar 6 16:09 UTC 2007 |
resp: #7: I just read that back. Please forgive my utter mangling of the
English language.
I guess I wasn't differentiating between the two different types of
FSCKs. In my defense, I've not dicked around with Solaris in a while and
have been stuck in Linux-land.
|
mickeyd
|
|
response 10 of 10:
|
Mar 9 02:47 UTC 2007 |
Thanks! I was just curious. Ive seen the system come up and run fsck
automatically after losing power and not gracefully shutting it down, etc.
Just wasnt sure if there was any value in fsck'ing like every 92nd day or
something; apparently not. Thanks for all the replies!
|