1 |
Oliver Schad wrote: |
2 |
> Am Mittwoch, 21. Mai 2008 schrieb mir Edward Muller: |
3 |
> |
4 |
>> reiserfs, purely from the recovery angle. |
5 |
>> |
6 |
> |
7 |
> Yeah, if you have fuck up in the B-Trees, it's the best filesystem purely from |
8 |
> the rubbish heap's angle. |
9 |
> |
10 |
> regards |
11 |
> Oli |
12 |
> |
13 |
|
14 |
Oli is correct. If reiser dies, the data is completely lost. It writes |
15 |
to the journal first, then writes the data. That, and when it decides to |
16 |
completely kill your B-Trees, you're screwed. After three unrecoverable |
17 |
reiserfs issues, I moved over to ext3 and have been very happy. |
18 |
|
19 |
I personally think the speed differences on most production servers is |
20 |
negligible. At the end of the day, I'd much rather have my data intact |
21 |
than have it be X% faster in certain situations. Also, ext3 has quite a |
22 |
few configuration options to optimize for your particular needs. See |
23 |
/etc/mke2fs.conf for some example configurations. The main problem |
24 |
you'll have with ext3 is that you cannot change to things like the block |
25 |
size or number of inodes on-the-fly like you can with xfs. So make sure |
26 |
what you format with will suit your needs. The one other major strength |
27 |
of ext3 is that it is able to change to ext2 or ext4 without |
28 |
reformatting the partition. You cannot do this with reiser or any of the |
29 |
others. reiserfs3 is not compatible with reiserfs4 for example. |
30 |
|
31 |
And I've got to think that the NTFS comment was entirely sarcasm. |
32 |
|
33 |
Wendall |
34 |
-- |
35 |
gentoo-server@l.g.o mailing list |