REISER versus EXT4 on Suse 11.2
in [Suse]
|
Prev: Compiz
Next: Confused about a software update
From: Kevin Nathan on 18 Feb 2010 16:20 On Wed, 17 Feb 2010 21:45:01 +0000 David Bolt <blacklist-me(a)davjam.org> wrote: >[0] you may use jfs instead but, since I've never used jfs or looked >into its strengths and weaknesses, I don't know whether it would be >more suitable. > I am not impressed with JFS. We had it on about 60 client machines and, clients being clients, it was not unusual for them to shut down the computer via the power switch instead of the shut down menu item. Of course, this caused JFS to try an fsck at boot and it would, quite often, error out and reboot, forever -- until manually booted into runlevel 1 to run fsck manually. That problem went away when we upgraded each machine and used ext3, instead... :-) -- Kevin Nathan (Arizona, USA) Linux Potpourri and a.o.l.s. FAQ -- (temporarily offline) Open standards. Open source. Open minds. The command line is the front line. Linux 2.6.31.8-0.1-default 14:14pm up 22 days 1:14, 16 users, load average: 0.21, 0.24, 0.18
From: Eef Hartman on 22 Feb 2010 08:36 David Bolt <blacklist-me(a)davjam.org> wrote: > I found that there wasn't much noticeable difference between ext3 and > reiserfs in use. The ext3 file system was faster to mount but, when the > number of files inside directories increased past several thousand, > reiserfs became noticeably quicker at getting the directory listings. Unless you set the "dir_index" attribute ON the ext3 filesystem (older mke2fs versions didn't enable this by default), then it is "as fast". Ext3 still doesn't have the "shrink directory when files are removed from it" feature that reiserfs has got, but apart from that ext3/4 and reiserfs are comparable. And - of course - reiserfs isn't being maintained anymore! This is from my /etc/mke2fs.conf (openSUSE 11.1): [defaults] base_features = sparse_super,filetype,resize_inode,dir_index,ext_attr blocksize = 4096 inode_size = 256 inode_ratio = 16384 so these are the defaults it uses when creating a ext3 file system. -- ******************************************************************* ** Eef Hartman, Delft University of Technology, dept. SSC/ICT ** ** e-mail: E.J.M.Hartman(a)tudelft.nl - phone: +31-15-278 82525 ** *******************************************************************
From: JT on 22 Feb 2010 10:14 On 22/02/10 14:36, Eef Hartman wrote: > David Bolt <blacklist-me(a)davjam.org> wrote: > >> I found that there wasn't much noticeable difference between ext3 and >> reiserfs in use. The ext3 file system was faster to mount but, when the >> number of files inside directories increased past several thousand, >> reiserfs became noticeably quicker at getting the directory listings. >> > Unless you set the "dir_index" attribute ON the ext3 filesystem (older > mke2fs versions didn't enable this by default), then it is "as fast". > Ext3 still doesn't have the "shrink directory when files are removed > from it" feature that reiserfs has got, but apart from that ext3/4 and > reiserfs are comparable. > And - of course - reiserfs isn't being maintained anymore! > > This is from my /etc/mke2fs.conf (openSUSE 11.1): > [defaults] > base_features = sparse_super,filetype,resize_inode,dir_index,ext_attr > blocksize = 4096 > inode_size = 256 > inode_ratio = 16384 > > so these are the defaults it uses when creating a ext3 file system. > Good tip, the dir_index thingy! Thanks. There could, by the way, be a rationale for not 'shrinking' directories after deletion of files. Directories have a 'typical use' mostly. And when one gets filled up with files and subsequently emptied, it might cost more to do the allocate/shrink cycle each time than to just leave the space for future use. Besides that : if they do grow out of control, you could always do a mv(dir)-mkdir-mv(files) trick to reclaim space. -- Kind regards, JT
From: David Bolt on 22 Feb 2010 10:57 On Monday 22 Feb 2010 13:36, while playing with a tin of spray paint, Eef Hartman painted this mural: > David Bolt <blacklist-me(a)davjam.org> wrote: >> I found that there wasn't much noticeable difference between ext3 and >> reiserfs in use. The ext3 file system was faster to mount but, when the >> number of files inside directories increased past several thousand, >> reiserfs became noticeably quicker at getting the directory listings. > > Unless you set the "dir_index" attribute ON the ext3 filesystem (older > mke2fs versions didn't enable this by default), The last time I checked was with a pre-10.3 system, possibly 10.0 or even 9.3, and ext3 was as sluggish as hell with huge numbers of files in a directory. As to when the dir_index was added, I don't know. My old 9.1 system doesn't have an /etc/mke2fs.conf file, and the next oldest is a 10.3 system where it does exist and dir_index is included as a default. > And - of course - reiserfs isn't being maintained anymore! True. Still, I've not had any issues with it so, unless there's suddenly a major bug that no-one is going to touch, I'm going to carry on using it. Regards, David Bolt -- Team Acorn: www.distributed.net OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s openSUSE 11.0 32b | | | openSUSE 11.3M1 32b openSUSE 11.0 64b | openSUSE 11.1 64b | openSUSE 11.2 64b | TOS 4.02 | openSUSE 11.1 PPC | RISC OS 4.02 | RISC OS 3.11
From: David Bolt on 22 Feb 2010 11:01
On Monday 22 Feb 2010 13:36, while playing with a tin of spray paint, Eef Hartman painted this mural: Missed this bit in my last reply: > Ext3 still doesn't have the "shrink directory when files are removed > from it" feature that reiserfs has got As a F(generic)YI, although you most likely already know, XFS is another file system that shrinks the directories when files are removed. Regards, David Bolt -- Team Acorn: www.distributed.net OGR-NG @ ~100Mnodes RC5-72 @ ~1Mkeys/s openSUSE 11.0 32b | | | openSUSE 11.3M1 32b openSUSE 11.0 64b | openSUSE 11.1 64b | openSUSE 11.2 64b | TOS 4.02 | openSUSE 11.1 PPC | RISC OS 4.02 | RISC OS 3.11 |