[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: ReiserFS on blank hard drive



On Wed, May 16, 2001 at 01:27:49AM +0200, Philipp von Weitershausen wrote:
> 
> Okay, I didn't make it clear enough: I have a Pismo with the original 6 gig 
> hard drive. Now, having to have both operating systems installed, that is way 
> to small for the operations that I do. Thus I need a bigger drive _instead_ 
> of the 6 gig one. My question is: how would I get reiserfs partitions on the 
> drive and then, how would I get my system on them?

install both disks and use cpio.  i never reinstall the OS for
something as mundane as a disk upgrade. 

> But, looking at the on-going discussion, it doesn't seem to be wise using 
> reiserfs. I can partially understand Ethan's `but you have to wait for fsck 
> when you pull the plug' whine problem, but sometimes, my Pismo doesn't wakeup 
> from a snooze, thus, it'll fsck the next boot and that costs battery power. I 
> don't want to think about a 20 gig hard drive being fsck'ed...

try better partitioning.  i have /usr and /usr/local on thier own
partitions, always mounted readonly.  i have tought apt how to
remount /usr read-write while installing packages and then readonly
again when its done.  /usr is the worst offender for long fsck times
since it generally has the most files of any other filesystem (except
possibly /home depending on how much stuff you have).  

> Is there really no other way? What are these developers doing all day? 
> Shouldn't they be fixing the bugs in reiser or porting ext3 to kernel 2.4? 
> Somebody give those people a job! (just kidding, you know...)

fixing bugs takes time.  unfortunatly rather then fix endianess
issues, the reiserfs people thought thier time would better be spent
accusing alan cox of conspiring to get ext3 into the kernel first just
because ext3 is being worked on by a redhat employee.  (alan cox
stated some things he felt must be fixed before reiser would be ready
for kernel inclusion, the two main ones were: 64 bit clean, and endian
clean). 

> I guess I'll stay with ext2 then. /bin/ls in my source code is really no risk 
> that I wanna take.

you are looking for more reliability and stability, that is not
something you can really expect from any of the journalling
filesystems quite yet.  yes yes some people love to say IWFM, but
there are plenty of others with horror stories too. 

i would suggest splitting up your partitioning.  that is really the
best thing you can do.  your / should be 64MB, have a seperate /tmp,
/usr, /var, /home and maybe /usr/local if you use it alot.  keep /usr
(and /usr/local if its seperate) mounted readonly at all times, then
it won't need to be fscked on a bad boot.  that will save the majority
of time.  keep in mind that a 10GB filesystem that is only 10% full
won't take any time at all to fsck.  but a single bloated / partition
will take forever mainly because of /usr.  

-- 
Ethan Benson
http://www.alaska.net/~erbenson/

Attachment: pgp9K2ozQHsh1.pgp
Description: PGP signature


Reply to: