Re: cold laptops are no good & reiser rules


Subject: Re: cold laptops are no good & reiser rules
From: Christopher E. Brown (cbrown@woods.net)
Date: Sun Mar 09 2003 - 11:36:07 AKST


On 8 Mar 2003, Damien Hull wrote:

> Had a bit of a scare this morning. Thought I'd share my experience so no
> one else goes through the same thing.
>
> I forgot my laptop in the car last night so it was a little cold when I
> hit the on button. It turns out that cold laptops don't work well. After
> opening a few applications the drive started making noise and the system
> froze. First thing that popped into my head was that the drive was going
> bad.
>
> During a reboot it tells me that somethings wrong with the drive and
> stops. Again I'm thinking that my drive just died on me.

Yes, drives have a specific temp range they are happy in. You should
always allow a system to warm to above freezing before powerup. Below
freezing the lube is thick, and the tolerances are off (different
metals and different expansion/contraction rates, on the hot side
drive parts tend to fail before it becomes a major issue, but on the
cold side it can be a major issue), also in more extreme cold many of
the alloys used can be very brittle.

> After a couple of reboots the screen looks brighter and everything
> works. Guess it needed to worm up a bit. From now on I'm only booting
> worm laptops.
>
> Thank God for reiser fs. Without that I'd be reinstalling Debian.

What does reiser have to do with anything? Filesystems featuring fs
meta-data journaling make cleanup faster and keep the tree consisant,
therefore reducing the need for full fs checks, but in the event of a
major failure, or a power loss they provide no more protection against
data loss than any other good fs, including ext2.

A full fs meta-data *and* data journaling system can provide
protection of the tree and your data, but it requires awareness of
journaling systems and atomic operations within the filesystem, the
kernel *and* the applications. Other than old Tandems and other HA
big iron this is not something you will find.

It is slightly more likely that you will lose the primary superblock
to corruption on a powerfail during large write operation with ext2
than with ext3 or reiser, but it is only the primary, all ext
filesystems have at least 2 superblocks present (for just this
reason).

Not that I don't like ext3 and reiser, specially when dealing with
very large storage arrays it makes cleanup alot faster, and the
directory tree indexing present in both can speed up certain
operations a great deal. Greater data integrity protection however,
is not one of the benifits whatever the journaling fanatics may claim.
If you are in the middle of a multi-block disk write updating a file
and you lose power, you will probably still have a file of that name
present with the correct fs attributes, the data within it will still
be toast though.

-- 
I route, therefore you are.

--------- To unsubscribe, send email to <aklug-request@aklug.org> with 'unsubscribe' in the message body.



This archive was generated by hypermail 2a23 : Sun Mar 09 2003 - 11:35:41 AKST