| View previous topic :: View next topic |
| Author |
Message |
Dont Panic
Guru
Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.
|
 Posted: Tue Nov 02, 2010 10:47 pm Post subject: |
 |
|
| Enlight wrote: | | Second it advantages fs who rather to lay things as they come instead of spreading them to prevent files and directory fragmentation later. |
Evaluating the effect of fragmentation in Btrfs is going to be interesting. Btrfs currently has problems with fragmentation on things like MySQL databases. With their COW (Copy On Write) architecture, every time you touch a byte in a database, it causes a section of the file to be relocated on the COW.
You can mount a volume with a 'nodatacow' option, but since Btrfs is not journaled, you're back in the old ext2 days with respect to data integrity on an unclean shutdown (maybe the old ext2 days were't really that bad after all  ).
Since the current btrfsck will only check the system, and can't fix anything yet, you can get in trouble going with nodatacow.
You do have a defragmentation tool, but I'm told it can't be run recursively, and will only work on the directory it is directly pointed at. This is due to Btrfs' snapshot capability. When you take a snapshot of a directory, only the updated pieces of the file are re-written from then on. So a defrag tool should only be targetted at either the original or the copy. But you can't defrag both unless you make a new copy of the file (which defeats the cool space-saving features of Btrfs snapshots).
It makes sense to me, but that shortcoming really handicaps the usefulness of the online defrag feature. If you don't have some idea of where you are experiencing fragmentation, the defrag tool isn't going to help much. |
|
| Back to top |
|
 |
Kollin
Veteran
Joined: 25 Feb 2006
Posts: 1139
Location: Sofia/Bulgaria
|
| Posted: Wed Nov 03, 2010 12:53 pm Post subject: |
|
|
It seems I can`t find any info about when btrfs will be production ready 
_________________
"Dear Enemy: may the Lord hate you and all your kind, may you be turned orange in hue, and may your head fall off at an awkward moment."
"Linux is like a wigwam - no windows, no gates, apache inside..." |
|
| Back to top |
|
|
Dont Panic
Guru
Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.
|
| Posted: Wed Nov 03, 2010 7:58 pm Post subject: |
|
|
| Kollin wrote: | | It seems I can`t find any info about when btrfs will be production ready |
It depends on your definition of "production ready".
Some brave souls are already employing the Btrfs filesystem for important "production-like" roles.
However, the code base is still under heavy development, and will remain so for some time.
Although they make an effort to insure the development won't impact the stability of the code, there is still the possibility that pace of change will cause problems.
For my part, I'd like to see the code base be "feature complete", and for the level of patches and changes to be down to a low roar.
They still have important features to implement, such as a working btrfsck program and RAID 5/6. Although you may not care about RAID 5/6, it's a considerable change that always has the potential to impact other things. |
|
| Back to top |
|
|
petrjanda
Veteran
Joined: 05 Sep 2003
Posts: 1557
Location: Brno, Czech Republic
|
| Posted: Thu Nov 04, 2010 12:54 am Post subject: |
|
|
| Dont Panic wrote: |
They still have important features to implement, such as a working btrfsck program and RAID 5/6. Although you may not care about RAID 5/6, it's a considerable change that always has the potential to impact other things. |
In my opinion, any RAID level other than 1 should always be implemented in hardware. Software RAID and being "production-ready" kind of doesn't add up in my head.
_________________
There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
| Back to top |
|
|
devsk
Advocate
Joined: 24 Oct 2003
Posts: 2995
Location: Bay Area, CA
|
| Posted: Thu Nov 04, 2010 1:21 am Post subject: |
|
|
| petrjanda wrote: | | Dont Panic wrote: |
They still have important features to implement, such as a working btrfsck program and RAID 5/6. Although you may not care about RAID 5/6, it's a considerable change that always has the potential to impact other things. |
In my opinion, any RAID level other than 1 should always be implemented in hardware. Software RAID and being "production-ready" kind of doesn't add up in my head. |
Then, u haven't tried ZFS's RAID support.... |
|
| Back to top |
|
|
petrjanda
Veteran
Joined: 05 Sep 2003
Posts: 1557
Location: Brno, Czech Republic
|
| Posted: Thu Nov 04, 2010 3:04 am Post subject: |
|
|
| devsk wrote: | | petrjanda wrote: | | Dont Panic wrote: |
They still have important features to implement, such as a working btrfsck program and RAID 5/6. Although you may not care about RAID 5/6, it's a considerable change that always has the potential to impact other things. |
In my opinion, any RAID level other than 1 should always be implemented in hardware. Software RAID and being "production-ready" kind of doesn't add up in my head. |
Then, u haven't tried ZFS's RAID support.... |
You're right, I haven't(although I've used ZFS on Solaris briefly in production, thank god no more...odd things happening with it's snapshot system), but I've tried plenty other SoftRAID implementations in non-production situation and they are not something I'd put my life on the line for.
_________________
There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
| Back to top |
|
|
idella4
Retired Dev
Joined: 09 Jun 2006
Posts: 1600
Location: Australia, Perth
|
| Posted: Sat Nov 06, 2010 11:41 am Post subject: |
|
|
| Quote: |
Since the current btrfsck will only check the system, and can't fix anything yet, you can get in trouble going with nodatacow.
|
I have read a bit about btrfs and apparently it is flagged as the next generation of the ext family.
I currently have a post re retrieving data from a corrupted btrfsck saved in an image file.
I find it completely unexpected and completely pathetic that after a number of years of development of btrfs that its best efforts are a complete sham.
It can't correctly read the full size of a system.
It is easily broken and brought to its knees with nothing more than copy and delete.
The btrfsck doesn't repair anything that's broken making it not [at least less than] a btrfsck.
etc etc etc.
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses?
_________________
idella4@aus |
|
| Back to top |
|
|
Kollin
Veteran
Joined: 25 Feb 2006
Posts: 1139
Location: Sofia/Bulgaria
|
| Posted: Sun Nov 07, 2010 9:47 am Post subject: |
|
|
All this does not sound very optimistic
_________________
"Dear Enemy: may the Lord hate you and all your kind, may you be turned orange in hue, and may your head fall off at an awkward moment."
"Linux is like a wigwam - no windows, no gates, apache inside..." |
|
| Back to top |
|
|
Ant P.
Watchman
Joined: 18 Apr 2009
Posts: 6920
|
| Posted: Sun Nov 07, 2010 11:09 am Post subject: |
|
|
| idella4 wrote: |
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses? |
Depends, how much are you paying them? |
|
| Back to top |
|
|
petrjanda
Veteran
Joined: 05 Sep 2003
Posts: 1557
Location: Brno, Czech Republic
|
| Posted: Sun Nov 07, 2010 12:31 pm Post subject: |
|
|
| Ant_P wrote: | | idella4 wrote: |
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses? |
Depends, how much are you paying them? |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
A good open source FS developed entirely without commercial backing checkout: HAMMER
_________________
There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
| Back to top |
|
|
devsk
Advocate
Joined: 24 Oct 2003
Posts: 2995
Location: Bay Area, CA
|
| Posted: Sun Nov 07, 2010 4:58 pm Post subject: |
|
|
| petrjanda wrote: | | Ant_P wrote: | | idella4 wrote: |
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses? |
Depends, how much are you paying them? |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
A good open source FS developed entirely without commercial backing checkout: HAMMER |
BTRFS is backed by Oracle, so your point is wrong. |
|
| Back to top |
|
|
kernelOfTruth
Watchman
Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)
|
| Posted: Sun Nov 07, 2010 5:12 pm Post subject: |
|
|
| petrjanda wrote: | | Ant_P wrote: | | idella4 wrote: |
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses? |
Depends, how much are you paying them? |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
A good open source FS developed entirely without commercial backing checkout: HAMMER |
unfortunately there doesn't seem to be any further porting work going on atm:
http://dlorch.github.com/hammer-linux/
at least linux-users have the ability to read hammer-filesystems
_________________
https://github.com/kernelOfTruth/ZFS-for-SystemRescueCD/tree/ZFS-for-SysRescCD-4.9.0
https://github.com/kernelOfTruth/pulseaudio-equalizer-ladspa
Hardcore Gentoo Linux user since 2004  |
|
| Back to top |
|
|
Dont Panic
Guru
Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.
|
| Posted: Sun Nov 07, 2010 6:09 pm Post subject: |
|
|
| devsk wrote: | | petrjanda wrote: | I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
A good open source FS developed entirely without commercial backing checkout: HAMMER |
BTRFS is backed by Oracle, so your point is wrong. |
Not only is Btrfs backed by Oracle, but I see significant contributers from people at Fujitsu, HP, and RedHat, to name just a couple.
One problem Btrfs has had recently is that it went about four to five months with nothing happening. I'm not 100% sure why this was going on, but very little new development was put in the kernel for the 2.6.35 and 2.6.36 kernels.
Currently, things are moving forward again.
I'm hopeful that many of the perceptions people are talking about in these last few posts are related to this gap in progress.
But I think it's fair to put Btrfs on your watch-list as a project that was expected to be further along by now. |
|
| Back to top |
|
|
petrjanda
Veteran
Joined: 05 Sep 2003
Posts: 1557
Location: Brno, Czech Republic
|
| Posted: Mon Nov 08, 2010 1:30 am Post subject: |
|
|
| devsk wrote: | | petrjanda wrote: | | Ant_P wrote: | | idella4 wrote: |
In short, it's an incompetent failure. How can qualified devs work on such a file system for years and produce such a list of serious misses? |
Depends, how much are you paying them? |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
A good open source FS developed entirely without commercial backing checkout: HAMMER |
BTRFS is backed by Oracle, so your point is wrong. |
I'm not sure if that invalidates or enhances what I said .
_________________
There is, a not-born, a not-become, a not-made, a not-compounded. If that unborn, not-become, not-made, not-compounded were not, there would be no escape from this here that is born, become, made and compounded. - Gautama Siddharta |
|
| Back to top |
|
|
idella4
Retired Dev
Joined: 09 Jun 2006
Posts: 1600
Location: Australia, Perth
|
| Posted: Tue Nov 09, 2010 3:40 pm Post subject: |
|
|
| petrjanda wrote: |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
|
I followed a post by devsk as he was testing btrfs and that should be well known by all. devsk's post was a type of into into the btrfs detail.
For fun and interest I am tinkering with it, and devsk you ignored the IM I sent you. Fell back to the reliable NeddySeagoon to reply.
Out of interest I re-ran the test of that post on reiserfs4 and as devsk has pointed it he couldn't break it, and I found it did all the right things.
Just bought a magazine today which has suse-11.3 on the dvd. On reading it, the new suse-11.3 features the new btrfs.
Well, that should be interesting.
| Dont Panic wrote: |
But I think it's fair to put Btrfs on your watch-list as a project that was expected to be further along by now
|
Another source says that btrfs is already flagged as the next generation of ext based filesystems for linux, then I find suse has moved.
It's an unreliable fragile under developed experiment. sheesh. Suse must know what they are unleashing!!??!!
_________________
idella4@aus |
|
| Back to top |
|
|
devsk
Advocate
Joined: 24 Oct 2003
Posts: 2995
Location: Bay Area, CA
|
| Posted: Tue Nov 09, 2010 5:28 pm Post subject: |
|
|
| idella4 wrote: | | petrjanda wrote: |
I think it's safe to say that if btrfs becomes a failure(personally I hope not) in the months/years to come that Linux devs aren't good file system designers. JFS,XFS,ZFS all were contributed by a company, ext2 is based on FFS. The only exceptions is ReiserFS who's developer sits in jail, and Reiser4 that got repeatedly rejected by kernel devs because of being "too different".
|
I followed a post by devsk as he was testing btrfs and that should be well known by all. devsk's post was a type of into into the btrfs detail.
For fun and interest I am tinkering with it, and devsk you ignored the IM I sent you. Fell back to the reliable NeddySeagoon to reply.
Out of interest I re-ran the test of that post on reiserfs4 and as devsk has pointed it he couldn't break it, and I found it did all the right things.
Just bought a magazine today which has suse-11.3 on the dvd. On reading it, the new suse-11.3 features the new btrfs.
Well, that should be interesting.
| Dont Panic wrote: |
But I think it's fair to put Btrfs on your watch-list as a project that was expected to be further along by now
|
Another source says that btrfs is already flagged as the next generation of ext based filesystems for linux, then I find suse has moved.
It's an unreliable fragile under developed experiment. sheesh. Suse must know what they are unleashing!!??!! |
idella4, my apologies for not responding to you. I have no idea how I missed the pm you sent me. I never got any email saying I have a PM. anyway, I will read it now. |
|
| Back to top |
|
|
idella4
Retired Dev
Joined: 09 Jun 2006
Posts: 1600
Location: Australia, Perth
|
| Posted: Tue Nov 09, 2010 6:06 pm Post subject: |
|
|
devsk;
ok no worries, you have a new IM and the link. I kind of got there in the end. It was just another gentoo linux learning exercise.
Hmmmmm, in case the IM has a problem, just as well enter it here
_________________
idella4@aus |
|
| Back to top |
|
|
Alex Libman
n00b
Joined: 26 May 2010
Posts: 39
Location: Leaving New Hampshire
|
| Posted: Mon Nov 22, 2010 3:17 am Post subject: |
|
|
| Btrfs has some nice features, but it is crippled by default due to restrictive licensing. What I would really like to see is HAMMER filesystem from DragonFly BSD ported to everything in sight -- Loonix, Winblows, MucOS, other BSD's, Solaris, AIX, VMware ESX(i), Haiku OS, etc, etc, etc -- which it can be due to its permissive license, and become the first truly universal ultra-modern FS. This would make it something that no GPL-licensed system can ever achieve - an industry standard, with high-quality proprietary tools like NTFS has pushing innovation forward. Its multi-master clustering and other advanced features would make it the ideal cloud storage standard as well. |
|
| Back to top |
|
|
soya
Guru
Joined: 29 Jan 2004
Posts: 429
|
| Posted: Tue Nov 23, 2010 9:53 am Post subject: |
|
|
Hi, i'mplanning to put my /usr/portage (without /usr/portage/distfiles) on btrfs for better performance, could someone tell me which are the best performance options for mkfs and /etc/fstab?
Thanks in advanced!
_________________
Si el tonto te alaba laméntalo, si el listo te censura piénsalo. |
|
| Back to top |
|
|
ToeiRei
Veteran
Joined: 03 Jan 2005
Posts: 1191
Location: Austria
|
| Posted: Tue Nov 23, 2010 1:55 pm Post subject: |
|
|
compression is nice here as a speed gain. Using it that way since some time.
_________________
Please stand by - The mailer daemon is busy burning your messages in hell... |
|
| Back to top |
|
|
gringo
Advocate
Joined: 27 Apr 2003
Posts: 3793
|
| Posted: Fri Dec 03, 2010 2:43 pm Post subject: |
|
|
talking about compression, has anyone played around with lzo compression in btrfs ?
i will probably play with it this weekend, was just wondering if anyone had some troubles with it already 
cheers ! |
|
| Back to top |
|
|
Dont Panic
Guru
Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.
|
| Posted: Sat Dec 04, 2010 3:23 am Post subject: |
|
|
| gringo wrote: | | talking about compression, has anyone played around with lzo compression in btrfs ? |
I've patched my kernel to support lzo compression, and it's been working fine for me, so far. It works pretty much as expected. Speed is faster, but compression is less.
I've used it on a secondary Gentoo partition, so I can't say I'm using lzo on a day-in-day-out basis. But I ran my secondary partition through the normal Gentoo workflow (a few dozen emerges and similar operations), and didn't encounter any problems.
It's difficult to say how much testing these patches have seen. Since the 2.6.37_rc process is too far along for a new feature like this, they don't plan on attempting to roll it out until the 2.6.38_rc period. |
|
| Back to top |
|
|
soya
Guru
Joined: 29 Jan 2004
Posts: 429
|
| Posted: Thu Dec 09, 2010 5:39 pm Post subject: |
|
|
Could you post a link to get lzo patch? thx
_________________
Si el tonto te alaba laméntalo, si el listo te censura piénsalo. |
|
| Back to top |
|
|
ToeiRei
Veteran
Joined: 03 Jan 2005
Posts: 1191
Location: Austria
|
| Posted: Thu Dec 09, 2010 6:46 pm Post subject: |
|
|
there is a feature in .37(-rc*) so far called free-space caching which should be even more interesting as it should make writing faster on btrfs. Maybe worth watching it.
_________________
Please stand by - The mailer daemon is busy burning your messages in hell... |
|
| Back to top |
|
|
Dont Panic
Guru
Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.
|
| Posted: Thu Dec 09, 2010 7:22 pm Post subject: |
|
|
| soya wrote: | | Could you post a link to get lzo patch? thx |
Li Zefan is the developer who has written the lzo patches.
He has put up a public repository containing his latest btrfs proposals at http://repo.or.cz/w/linux-btrfs-devel.git
For myself, I usually keep an eye on the Mailing List, and pluck selected patches of interest from LKML's patchwork page for Btrfs. I click on a patch of interest, then click on the mbox link to download the patch as a "git format-patch" patch. I usually then apply these to a local git kernel repository I have to play with these git patches.
| ToeiRei wrote: | | there is a feature in .37(-rc*) so far called free-space caching which should be even more interesting as it should make writing faster on btrfs. Maybe worth watching it. |
I've done some benchmarking while testing these patches. I was unable to find any speed improvement while testing with space_cache enabled and disabled. I swapped some emails with the developer of these patches, and he indicated that these patches should not affect performance too much.
I'd be curious if anybody else has a different experience. |
|
| Back to top |
|
|
|
|
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
|
|
Gentoo Chat
