Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Some ext3 Filesystem Tips
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3 ... 13, 14, 15  
Reply to topic    Gentoo Forums Forum Index Documentation, Tips & Tricks
View previous topic :: View next topic  
Author Message
JeliJami
Veteran
Veteran


Joined: 17 Jan 2006
Posts: 1086
Location: Belgium

PostPosted: Mon Jan 14, 2008 10:20 am    Post subject: Reply with quote

Schizoid wrote:
I would think that if there was some lost data it found that it would recreate that directory as needed?

No it doesn't. If your lucky, it will restore a complete file into the lost+found directory, but with some predefined name; FSCK00001 for example, don't remember exactly (DOS's checkdisk utility did something similar, with CHCKDSK.001, I think)
But most of the time, you will only get partial files, without any clue to the original filename, or its original path.
_________________
Unanswered Post Initiative | Search | FAQ
Former username: davjel
Back to top
View user's profile Send private message
pactoo
Guru
Guru


Joined: 18 Jul 2004
Posts: 553

PostPosted: Thu Jan 17, 2008 6:52 am    Post subject: Reply with quote

>I don't know why would anyone want to delete that directory, though...

Probably because nobody (except those few 20+ years Unix experienced die hard ubergeeks) knows how to actually recover those files, that are put in there after and by fsck.
Back to top
View user's profile Send private message
number_nine
Tux's lil' helper
Tux's lil' helper


Joined: 05 May 2005
Posts: 136

PostPosted: Mon Mar 31, 2008 2:01 pm    Post subject: Reply with quote

Are these Ext3 tips still relevant?

I've done some testing on my computer with gentoo-sources-2.6.23-r9 and bonnie++ (v1.03 compiled from source).

At this point, I'm most concerned with iowait associated with write performance:

Code:

                      Version  1.03      ------Sequential Output------
                                         -Per Chr- --Block-- -Rewrite-
                            Machine Size K/sec %CP K/sec %CP K/sec %CP
ext3-defaults               gentoo    8G 54120  78 62062  15 28364   5
ext3-noatime,journal_data   gentoo    8G 22792  34 30598   8 19971   4
ext3-noatime,writeback_data gentoo    8G 59236  85 60514  13 27127   4
ext2-noatime                gentoo    8G 63030  84 65542   6 27931   4


Notice how when I use journal=data (as suggested in this thread), I have the least write performance, but also decreased CPU usage.

The best performance appears to be with ext2, followed by data=writeback with ext3.

Thoughts?
Back to top
View user's profile Send private message
XenoTerraCide
Veteran
Veteran


Joined: 18 Jan 2004
Posts: 1418
Location: MI, USA

PostPosted: Mon Mar 31, 2008 2:26 pm    Post subject: Reply with quote

write is slower with data=journal than writeback. however write is faster that writeback while reading. and of course ext2 is faster it doesn't have journaling, which means less overhead. FAT is probably faster 2. but I would use it because I love my data.
_________________
I don't hang out here anymore, try asking on http://unix.stackexchange.com/ if you want my help.
Back to top
View user's profile Send private message
XenoTerraCide
Veteran
Veteran


Joined: 18 Jan 2004
Posts: 1418
Location: MI, USA

PostPosted: Thu May 01, 2008 9:10 pm    Post subject: Reply with quote

http://www.linuxplanet.com/linuxplanet/tutorials/6480/1/

interesting updates to ext3 that is causing problems with grub. perhaps something should be added to our tips? also does anyone know if these updates otherwise affect our tips. (such as whether data_journal is still the king of read)
_________________
I don't hang out here anymore, try asking on http://unix.stackexchange.com/ if you want my help.
Back to top
View user's profile Send private message
XenoTerraCide
Veteran
Veteran


Joined: 18 Jan 2004
Posts: 1418
Location: MI, USA

PostPosted: Mon Dec 29, 2008 7:14 am    Post subject: Reply with quote

anyone know if data_journal offers the same benefits in ext4 (faster reads while writing)? or if there are any new enhancements and tweaks we can make?
_________________
I don't hang out here anymore, try asking on http://unix.stackexchange.com/ if you want my help.
Back to top
View user's profile Send private message
arnuld
Tux's lil' helper
Tux's lil' helper


Joined: 13 Apr 2007
Posts: 124

PostPosted: Mon Feb 16, 2009 11:24 am    Post subject: Reply with quote

hey, what about noatime and reltime options ? I searched some places and and found that people were praising these 2 mount options:

http://kerneltrap.org/node/14148
http://www.pervasivecode.com/blog/2008/05/15/recommended-mount-options-for-ext3/
_________________
arnuld
http://lispmachine.wordpress.com
Back to top
View user's profile Send private message
XenoTerraCide
Veteran
Veteran


Joined: 18 Jan 2004
Posts: 1418
Location: MI, USA

PostPosted: Mon Feb 16, 2009 5:22 pm    Post subject: Reply with quote

yeah... but noatime isn't really ext(x) specific. That applies to everyfs and as far as I can tell there is never a reason not to use it. I've never had one.
_________________
I don't hang out here anymore, try asking on http://unix.stackexchange.com/ if you want my help.
Back to top
View user's profile Send private message
BlackB1rd
Tux's lil' helper
Tux's lil' helper


Joined: 13 Aug 2003
Posts: 131
Location: /Europe/Netherlands/Haarlem

PostPosted: Sat Mar 07, 2009 12:29 pm    Post subject: Reply with quote

Is it correct to assume that ext3 with data=journal performs better than default (ordered) on a server platform with multiple databases and "normal files" accessed by many concurrent users? The articles found on the internet are all written many years ago and I'm not sure if those results are still valid. And my biggest question would be why the default has changed from journal to ordered when the latter performs better on servers? Shouldn't that be the default setting when performing a server installation?
Back to top
View user's profile Send private message
Cyker
Veteran
Veteran


Joined: 15 Jun 2006
Posts: 1746

PostPosted: Sun Mar 08, 2009 10:01 am    Post subject: Reply with quote

BlackB1rd wrote:
Is it correct to assume that ext3 with data=journal performs better than default (ordered) on a server platform with multiple databases and "normal files" accessed by many concurrent users? The articles found on the internet are all written many years ago and I'm not sure if those results are still valid. And my biggest question would be why the default has changed from journal to ordered when the latter performs better on servers? Shouldn't that be the default setting when performing a server installation?


AFAIK, the default journal mode has always been data=ordered.

data=journal is only faster when the filesystem is having to do lots of reads AND writes at the same time.
Formostly-reads, the other two are faster.

I used to use data=journal, but had to go back to data=ordered as data=journal makes ext4 throw out some warnings about disabling some of its features.

Note 'tho, that 'faster' is not some order of magnitude thing; It's the sort of 'faster' that is only really noticable in benchmarks :)
Back to top
View user's profile Send private message
XenoTerraCide
Veteran
Veteran


Joined: 18 Jan 2004
Posts: 1418
Location: MI, USA

PostPosted: Sun Mar 08, 2009 1:44 pm    Post subject: Reply with quote

delayed allocation hasn't been written for data=journal yet in ext4, this is only a temporary thing. so far as I know the rest of the ext4 advantages work.
_________________
I don't hang out here anymore, try asking on http://unix.stackexchange.com/ if you want my help.
Back to top
View user's profile Send private message
Strowi
l33t
l33t


Joined: 19 Aug 2003
Posts: 656
Location: Bonn

PostPosted: Mon Jun 29, 2009 6:40 am    Post subject: stride on dmraid? Reply with quote

hi,

i've been looking into this thread from time to time... does anyone have a clue about stride size for dmraid raid0? I am an using an nvidia-dmraid 2hdds and 1 350GB ext3-partition. I chose 64kb as stripe-size in the controller-bios and "dmraid -s" reports a stride-size of 128, should i use 128 as stride-option for mkfs?
greetings and thx for all the tips,
_________________
--
Linux & such ...
http://blog.hasnoname.de
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Documentation, Tips & Tricks All times are GMT
Goto page Previous  1, 2, 3 ... 13, 14, 15
Page 15 of 15

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum