Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Reiser4 Gentoo FAQ [25Dec2016]
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
Dont Panic
Guru
Guru


Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.

PostPosted: Mon Jan 19, 2009 3:43 pm    Post subject: Reply with quote

kernelOfTruth wrote:
someone should really create a wiki where we can write down all of the important tid bits

Gentoo Forums...FTW!
Back to top
View user's profile Send private message
minor_prophets
Apprentice
Apprentice


Joined: 07 Oct 2007
Posts: 281

PostPosted: Thu Jan 22, 2009 8:08 pm    Post subject: Reply with quote

Ok. I'm away from my system, but I'm having trouble getting / to mount rw.
my /boot is ext2. I can post more later, but I'm missing an example of a grub rw line for grub.conf to try when I get back. Can someone please provide?

Thanks all.
Back to top
View user's profile Send private message
minor_prophets
Apprentice
Apprentice


Joined: 07 Oct 2007
Posts: 281

PostPosted: Thu Jan 22, 2009 8:32 pm    Post subject: Reply with quote

adding rw after my root=/dev/sda3 line like so?
Code:
...root=/dev/sda3 rw
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Thu Jan 22, 2009 10:54 pm    Post subject: Reply with quote

minor_prophets wrote:
adding rw after my root=/dev/sda3 line like so?
Code:
...root=/dev/sda3 rw


yeah, in that way:

Quote:
root (hd0,0)
kernel (hd0,0)/vmlinuz-2.6.29-rc2-zen1 root=/dev/sdc5 ro elevator=cfq


but why are you having problems with it not being remounted rw ?

openrc ?
_________________
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 :D
Back to top
View user's profile Send private message
minor_prophets
Apprentice
Apprentice


Joined: 07 Oct 2007
Posts: 281

PostPosted: Thu Jan 22, 2009 11:31 pm    Post subject: Reply with quote

No, I'm not sure yet. I get a variant of the following. By this I meant that the below error will occur at different times between various boot scripts. I've seen three variants of its occurrence across a few reboots. All culminating in either the (none) part below or a control-D, neither of which I'm able to run anything seemingly.

Code:
*Skipping /var and /tmp initialization (ro root?)
...
*Skipping /var/run/utmp initialization (ro root?)
/sbin/runscript:  error while loading shared libraries:  /lib/libdl.so.2:  invalid ELF header
INIT:  Entering runlevel:  3

then dropped to
Code:
This is (none).unknown_domain (Linux i686 2.6.25-hardened-r12)
(none) login:root
/sbin/runscript:  error while loading shared libraries:  /lib/libdl.so.2:  invalid ELF header


All partitions except /boot are reiser4. Booting(and installed) from systemrescue cd. I give you a peek at my fstab and grub.conf in a moment. I've been banging my head on this one now for a few days. Its gotta be some silly oversight somewhere, I'm thinking.

fstab
Code:
/dev/sda1               /boot           ext2            noauto,noatime  1 2
/dev/sda3               /               reiser4         noatime,nodiratime,flush,scan_maxnodes=15000    0 1
/dev/sda2               none            swap            sw              0 0
/dev/sda5               /opt            reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda6               /usr            reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda7               /usr/portage    reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda8               /usr/portage/distfiles  reiser4 noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda9               /var            reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda10              /var/tmp        reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda11              /tmp            reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0
/dev/sda12              /home           reiser4         noatime,nodiratime,flush,scan_maxnodes=1500     0 0

grub.conf
Code:
default 0
timeout 3
#splashimage=(hd0,0)/boot/grub/splash.xpm.gz

#title Gentoo Linux 2.6.24-r5
#root (hd0,0)
#kernel /boot/kernel-genkernel-x86-2.6.24-gentoo-r5 root=/dev/ram0 real_root=/dev/sda3
#initrd /boot/initramfs-genkernel-x86-2.6.24-gentoo-r5

# vim:ft=conf:

title Gentoo Linux 2.6.25-hardened-r12
root (hd0,0)
kernel /boot/kernel-2.6.25-hardened-r12 root=/dev/sda3

title Gentoo Linux 2.6.25-r12 nohardened
root (hd0,0)
kernel /boot/kernel-2.6.25-hardened-r12-nohardened root=/dev/sda3

Code:
RC_PARALLEL_STARTUP="yes"

This is a fresh installation and I've emerge -e world after setting a hardened profile with eselect. I used a stage3 hardened i686. Can't think of what else to tell you all at this point. :? Ask and I will do my best to provide the info.

PLEASE SEE AND REPLY HERE: https://forums.gentoo.org/viewtopic-t-728990-highlight-reiser4.html
That is the running thread I'd started. I don't want to cross-talk/hijack this thread. Thanks.
Back to top
View user's profile Send private message
minor_prophets
Apprentice
Apprentice


Joined: 07 Oct 2007
Posts: 281

PostPosted: Tue Jan 27, 2009 5:31 pm    Post subject: Reply with quote

Well, I abandoned r4 for r36. I have other machines for R4 experiments. Maybe an older Inpsiron laptop, a 2650 or an 8100.
Back to top
View user's profile Send private message
Dont Panic
Guru
Guru


Joined: 20 Jun 2007
Posts: 322
Location: SouthEast U.S.A.

PostPosted: Tue Jan 27, 2009 10:36 pm    Post subject: Re: Reiser4 Gentoo FAQ Reply with quote

I was looking over this FAQ after updating to sys-fs/reiser4progs-1.0.7

dusanc wrote:

Q4: I'm having trouble with openRC and Reiser4, how can I fix it?

Set fsck_args="-a" in /etc/conf.d/net


First, shouldn't that be /etc/conf.d/fsck and not /etc/conf.d/net?

But, now that reiser4progs has been updated, I've gone back to my /etc/conf.d/fsck and changed fsck_args back from "-a" to "-p" (which I think was the original default).

I am running sys-apps/openrc-0.2.5, and am able to boot now without problems with fsck_args set to "-p".

What is the difference between "-a" and "-p" for reiser4progs. Looking at the man page, "-a" "automatically checks the file system without any questions", and "-p" "automatically repairs minor corruptions on the filesystem". I'm not sure I appreciate the difference.

My openrc package is alittle old (there is a version 0.4.2 available), so I'll upgrade my openrc, and post back if I have any problems with other versions.
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 283

PostPosted: Mon Feb 09, 2009 10:48 am    Post subject: Data Security Reply with quote

I don't know if it is my setup and configuration or what, but each time I have a crash and I have any read/write activity on a reiser4 partition, I have data loss.

Here is what I've learned though from these crashes
    1) Any file being access, regardless of whether or not it is being modified or just read once, will become detached from the file system. Albeit, leaf, node, the whole tree.
    2) Reiser4 loves RAM and this is a double edges sword. After seeing files continually get tossed into lost+found when I have a system lock up on me, this became a sharp double-edged sword.
    3) Adding a cron task "sync" running every 5 minutes or so, I'd naturally avoid the above if the system was idle and not reading files on said partitions
    4) Make regular backups, daily, weekly, monthly, and incremental before/after major system updates.
    5) Reiser4 is listed as Experimental for a reason. Head this warning as everyone says.


That being said. I'm taking some time out of re-installing Gentoo again while I'm on a LiveCD to see what mounting options I can tweak to find out how to secure my data better but still enjoy the speed or Reiser4. My learnings will be posted over here
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://mobrienphotography.com/
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Mon Feb 09, 2009 3:31 pm    Post subject: Re: Data Security Reply with quote

DigitalCorpus wrote:
I don't know if it is my setup and configuration or what, but each time I have a crash and I have any read/write activity on a reiser4 partition, I have data loss.

Here is what I've learned though from these crashes
    1) Any file being access, regardless of whether or not it is being modified or just read once, will become detached from the file system. Albeit, leaf, node, the whole tree.
    2) Reiser4 loves RAM and this is a double edges sword. After seeing files continually get tossed into lost+found when I have a system lock up on me, this became a sharp double-edged sword.
    3) Adding a cron task "sync" running every 5 minutes or so, I'd naturally avoid the above if the system was idle and not reading files on said partitions
    4) Make regular backups, daily, weekly, monthly, and incremental before/after major system updates.
    5) Reiser4 is listed as Experimental for a reason. Head this warning as everyone says.


That being said. I'm taking some time out of re-installing Gentoo again while I'm on a LiveCD to see what mounting options I can tweak to find out how to secure my data better but still enjoy the speed or Reiser4. My learnings will be posted over here


I've made the exact same experience with reiser4, DigitalCorpus,

and that's also the reason why it (still) prevents me from using it on my /home partition

there's still the (very tiny) risk of losing valuable data I've been working on after an unforseeable crash - which doesn't happen with reiserfs

you mgiht want to tinker a little with the following VM settings:

/proc/sys/vm/dirty_expire_centisecs
/proc/sys/vm/dirty_writeback_centisecs <-- very important
/proc/sys/vm/dirty_background_ratio
/proc/sys/vm/dirty_ratio

also during the time I had reiser4 on my /home partition I did some sync commands manually to ensure data consistency / data is on the harddrive

I'll post more if I find out about how to disarm / lower the amount of data loss during a crash
_________________
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 :D
Back to top
View user's profile Send private message
darklegion
Guru
Guru


Joined: 14 Nov 2004
Posts: 468

PostPosted: Tue Feb 10, 2009 5:28 am    Post subject: Reply with quote

Are you guys using cryptcompress on the filesystems where you are getting data loss? I uses to have this problem with older reiser4 versions and stock reiser4 but it hasn't happened in many kernel releases, as long as cryptcompress is not enabled.With cryptcompress it still seems to happen, usually it only corrupts executables although recently I lost most of my system drive with cryptcompress enabled.So if you are using compression, that's probably what's causing it.I only use it on /usr/portage now because of this.

On a related note, are there any reiser4 recovery utilities yet? All of the stuff I lost is replaceable but I'd still like to know for future reference.I could just use magicrescue/foremost but losing filenames is a big deal for system files.
Back to top
View user's profile Send private message
dusanc
Apprentice
Apprentice


Joined: 19 Sep 2005
Posts: 248
Location: Serbia

PostPosted: Tue Feb 10, 2009 6:30 am    Post subject: Reply with quote

It would be really nice if someone could find a way to reproduce this problem.
We can't expect things to get fixed if we don't report them :)
_________________
Reiser4 Gentoo FAQ [25Dec2016]
Back to top
View user's profile Send private message
zyko
l33t
l33t


Joined: 01 Jun 2008
Posts: 620
Location: Munich, Germany

PostPosted: Tue Feb 10, 2009 9:47 am    Post subject: Reply with quote

Quote:
Are you guys using cryptcompress on the filesystems where you are getting data loss? I uses to have this problem with older reiser4 versions and stock reiser4 but it hasn't happened in many kernel releases, as long as cryptcompress is not enabled.With cryptcompress it still seems to happen, usually it only corrupts executables although recently I lost most of my system drive with cryptcompress enabled.So if you are using compression, that's probably what's causing it.


How exactly does this data loss manifest?

Reiser4 + Compression does have the annoying "false positive" bug when fscking it (=will display loads of errors), but the data is never actually corrupted.

Do you maybe have an example of a file that was corrupted by reiser4 + a original, uncorrupted backup of the same file?
Back to top
View user's profile Send private message
darklegion
Guru
Guru


Joined: 14 Nov 2004
Posts: 468

PostPosted: Wed Feb 11, 2009 5:37 am    Post subject: Reply with quote

zyko wrote:
Quote:
Are you guys using cryptcompress on the filesystems where you are getting data loss? I uses to have this problem with older reiser4 versions and stock reiser4 but it hasn't happened in many kernel releases, as long as cryptcompress is not enabled.With cryptcompress it still seems to happen, usually it only corrupts executables although recently I lost most of my system drive with cryptcompress enabled.So if you are using compression, that's probably what's causing it.


How exactly does this data loss manifest?

Reiser4 + Compression does have the annoying "false positive" bug when fscking it (=will display loads of errors), but the data is never actually corrupted.

Do you maybe have an example of a file that was corrupted by reiser4 + a original, uncorrupted backup of the same file?


Well, I'll try and explain what recently happened my case.First off, I was just using my PC normally when it started randomly becoming unstable and then locking up; reiser4 errors would appear in dmesg when this happened.In some cases, a reboot + fsck.reiser4 --build-fs would be enough to fix it, but in the most recent case, various executables were completely zeroed out (mostly in /bin).
The next issue was that fsck.reiser4 started segfaulting at the same place fsck.reiser4 --build-fs.The places where it would segfault were mostly where zeroed out files, and I was able to delete most of these.I was getting input/output errors in some directories though, so was not able to delete all of them.After another attempt at using fsck.reiser4 --build-fs to fix things, I mounted the filesystem, and was left with most of the data on my system drive lost, with some of it left in lost+found.
I thought that the input/output errors may have pointed to a hardware problem, but I backed up the filesystem with ddrescue, and there were no issues while copying it.I mounted the backup filesystem and the i/o errors were still there.
I should note that I use LUKS encryption on the system drive, so maybe reiser4 + cryptcompress has issues with this.Since other people have mentioned data loss, it seems more likely that it occurs even without encryption.
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Wed Feb 11, 2009 11:11 am    Post subject: Reply with quote

darklegion wrote:
zyko wrote:
Quote:
Are you guys using cryptcompress on the filesystems where you are getting data loss? I uses to have this problem with older reiser4 versions and stock reiser4 but it hasn't happened in many kernel releases, as long as cryptcompress is not enabled.With cryptcompress it still seems to happen, usually it only corrupts executables although recently I lost most of my system drive with cryptcompress enabled.So if you are using compression, that's probably what's causing it.


How exactly does this data loss manifest?

Reiser4 + Compression does have the annoying "false positive" bug when fscking it (=will display loads of errors), but the data is never actually corrupted.

Do you maybe have an example of a file that was corrupted by reiser4 + a original, uncorrupted backup of the same file?


Well, I'll try and explain what recently happened my case.First off, I was just using my PC normally when it started randomly becoming unstable and then locking up; reiser4 errors would appear in dmesg when this happened.In some cases, a reboot + fsck.reiser4 --build-fs would be enough to fix it, but in the most recent case, various executables were completely zeroed out (mostly in /bin).
The next issue was that fsck.reiser4 started segfaulting at the same place fsck.reiser4 --build-fs.The places where it would segfault were mostly where zeroed out files, and I was able to delete most of these.I was getting input/output errors in some directories though, so was not able to delete all of them.After another attempt at using fsck.reiser4 --build-fs to fix things, I mounted the filesystem, and was left with most of the data on my system drive lost, with some of it left in lost+found.
I thought that the input/output errors may have pointed to a hardware problem, but I backed up the filesystem with ddrescue, and there were no issues while copying it.I mounted the backup filesystem and the i/o errors were still there.
I should note that I use LUKS encryption on the system drive, so maybe reiser4 + cryptcompress has issues with this.Since other people have mentioned data loss, it seems more likely that it occurs even without encryption.


I wouldn't use reiser4 + cryptcompress and LUKS that's the worst combination I've made experience so far (all of my data partitions are encrypted so I inevitably had to use it with LUKS),

I don't know how much time you have right now and/or willing to experiment but there's a patch for dm floating around over at lkml.org which permits to use async writes on device-mapper partitions / devices of which the underlying drives are only 1 physical drive - maybe that fixes it and it's a problem specific with dm-crypt / device-mapper and reiser4 + cryptcompress

all in all reiser4 is pretty safe data-safety wise - the only limitations / problems so far are:

- use with LUKS-partitions
- cryptcompress partitions to cryptcompresss partitions write / read operations when creating a stage4 tarball triggers BUGs / pseudo-corruption (I suspect a faulty kernel(patch) since it doesn't always happen)

I'll add more if there's still something to cross my mind
_________________
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 :D
Back to top
View user's profile Send private message
darklegion
Guru
Guru


Joined: 14 Nov 2004
Posts: 468

PostPosted: Wed Feb 11, 2009 12:05 pm    Post subject: Reply with quote

Do you mean this?
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Fri Feb 13, 2009 11:10 pm    Post subject: Reply with quote

darklegion wrote:
Do you mean this?


most likely not - I'm still seeing messages about reiser4 switching to synchronous writes in dmesg

I've found it - it's:

Quote:
http://lkml.org/lkml/2008/2/15/125
([PATCH] Implement barrier support for single device DM devices)
_________________
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 :D
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 283

PostPosted: Sat Feb 14, 2009 7:06 am    Post subject: Reply with quote

zyko wrote:
Quote:
Are you guys using cryptcompress on the filesystems where you are getting data loss? I uses to have this problem with older reiser4 versions and stock reiser4 but it hasn't happened in many kernel releases, as long as cryptcompress is not enabled.With cryptcompress it still seems to happen, usually it only corrupts executables although recently I lost most of my system drive with cryptcompress enabled.So if you are using compression, that's probably what's causing it.


How exactly does this data loss manifest?

Reiser4 + Compression does have the annoying "false positive" bug when fscking it (=will display loads of errors), but the data is never actually corrupted.

Do you maybe have an example of a file that was corrupted by reiser4 + a original, uncorrupted backup of the same file?

For Reiser4 + cryptcompress the corruption is in the filesystem structure, not the actual files. If a file is being accessed, that is either read or written, and the kernel panics or power is lost or what have you, before everything is synced/flushed back to the drive, those files will be lost to the file system. Running fsck.reiser4 --fix might bring them back to lost+found, else the --build-fs will do that for you.

I'm struggling with linux/gentoo as it is and trying to learn something along. I've had time to observe software development before, so I know how particular bug reporting can be. Let alone I'm not sure how to proceed in doing so. If I had the time and knew the language, I would help fix these bugs, but I've opted away from CompSci in my education :/
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://mobrienphotography.com/
Back to top
View user's profile Send private message
neuron
Advocate
Advocate


Joined: 28 May 2002
Posts: 2371

PostPosted: Sun Feb 15, 2009 12:06 pm    Post subject: Reply with quote

anyone know of any work going into r4 + latency? It's my one big gripe with R4+cryptcompress atm, latency is quite bad.
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 283

PostPosted: Sun Feb 15, 2009 12:34 pm    Post subject: Reply with quote

How do you mean latency? type in a command and wait a second or two before it executes after you hit enter? listing a directory isn't instantaneous? Cannot access a large number of files as quickly as other filesystems?

What kernel version are you using? Do you have Preemption enlabled? Is it forced? What is the timer frequency you selected when you compiled the kernel? What block I/O scheduler are you using? how old is the hard drive you're using? what CPU and architecture are you using?

I'm using R4 with 2.6.28, no preemption (well this is a server for the most part), frequency set to 1000Hz, using CFQ since it behaves better than Anticipatory on Reiser4 (this is probably your gripe right here as I noticed a big difference when i switched). Also under disk load x86_64 still suffers from poor disk I/O whether the load is caused by disk access or CPU utilization. I'm even susceptible to it on a Q6700. I've taken measures to counter it by using XFS's agsize as a multiple of my throughput/sec. I know there are some scheduling tweaks posted by kernelOfTruth over in the tuning thread for ReiserFS/Reiser4. Properly formating your disk/partition (cluster size of 4K will have less latency but more overhead than a cluster size of 64K) with cryptcompress can make a sizable difference in performance and realtime usage. I'm still tuning and tweaking tmgr.atom_max_size and tmgr.atom_max_age to see where I can get some stability and fluidity in disk access.
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://mobrienphotography.com/
Back to top
View user's profile Send private message
neuron
Advocate
Advocate


Joined: 28 May 2002
Posts: 2371

PostPosted: Sun Feb 15, 2009 12:52 pm    Post subject: Reply with quote

DigitalCorpus wrote:
How do you mean latency? type in a command and wait a second or two before it executes after you hit enter? listing a directory isn't instantaneous? Cannot access a large number of files as quickly as other filesystems?


Latency under load, when I try to do software development while running a backup task for example (with ionice -c3 and nice 19)

DigitalCorpus wrote:
What kernel version are you using? Do you have Preemption enlabled? Is it forced? What is the timer frequency you selected when you compiled the kernel? What block I/O scheduler are you using? how old is the hard drive you're using? what CPU and architecture are you using?


2.6-29-rc5 atm, prempt options dont make much difference. Switching to low latency desktop from desktop kills io performance while not doing much to latency. 1000hz, cfq, fairly recent drives, hd performance is fine. And I'm on an intel quad core.

DigitalCorpus wrote:
I'm using R4 with 2.6.28, no preemption (well this is a server for the most part), frequency set to 1000Hz, using CFQ since it behaves better than Anticipatory on Reiser4 (this is probably your gripe right here as I noticed a big difference when i switched). Also under disk load x86_64 still suffers from poor disk I/O whether the load is caused by disk access or CPU utilization. I'm even susceptible to it on a Q6700. I've taken measures to counter it by using XFS's agsize as a multiple of my throughput/sec. I know there are some scheduling tweaks posted by kernelOfTruth over in the tuning thread for ReiserFS/Reiser4. Properly formating your disk/partition (cluster size of 4K will have less latency but more overhead than a cluster size of 64K) with cryptcompress can make a sizable difference in performance and realtime usage. I'm still tuning and tweaking tmgr.atom_max_size and tmgr.atom_max_age to see where I can get some stability and fluidity in disk access.


I've just used create=ccreg40,compress=lzo1, no changes to default cluster sizes etc. And no tweaks to any scheduler/fs.

I'm using device mapper encryption however, could be they dont mix all that well. I realized ext4+luks seems to have way less latency issues than r4+luks does on my system. With ext4 I can barely notice, with r4 my mouse occasionally stops moving.

Note that these are drives on the same system, but different partitions, so the results are not directly comparable.
Back to top
View user's profile Send private message
dusanc
Apprentice
Apprentice


Joined: 19 Sep 2005
Posts: 248
Location: Serbia

PostPosted: Sun Feb 15, 2009 9:57 pm    Post subject: Reply with quote

noatime in mount options?
_________________
Reiser4 Gentoo FAQ [25Dec2016]
Back to top
View user's profile Send private message
neuron
Advocate
Advocate


Joined: 28 May 2002
Posts: 2371

PostPosted: Sun Feb 15, 2009 11:09 pm    Post subject: Reply with quote

that I do have
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 283

PostPosted: Mon Feb 16, 2009 9:26 am    Post subject: Reply with quote

neuron wrote:
DigitalCorpus wrote:
How do you mean latency? type in a command and wait a second or two before it executes after you hit enter? listing a directory isn't instantaneous? Cannot access a large number of files as quickly as other filesystems?


Latency under load, when I try to do software development while running a backup task for example (with ionice -c3 and nice 19)

DigitalCorpus wrote:
What kernel version are you using? Do you have Preemption enlabled? Is it forced? What is the timer frequency you selected when you compiled the kernel? What block I/O scheduler are you using? how old is the hard drive you're using? what CPU and architecture are you using?


2.6-29-rc5 atm, prempt options dont make much difference. Switching to low latency desktop from desktop kills io performance while not doing much to latency. 1000hz, cfq, fairly recent drives, hd performance is fine. And I'm on an intel quad core.

Using a 64bit bit environment, even if it is multilib, there are issues with disk I/O under load no matter the filesystem. Reiser4 might be hit the hardest on this since it requires a bit more CPU load, especially with cryptcompress, than other filesystems. Even with a C2Q, see my sig, you are not immune.
neuron wrote:
DigitalCorpus wrote:
I'm using R4 with 2.6.28, no preemption (well this is a server for the most part), frequency set to 1000Hz, using CFQ since it behaves better than Anticipatory on Reiser4 (this is probably your gripe right here as I noticed a big difference when i switched). Also under disk load x86_64 still suffers from poor disk I/O whether the load is caused by disk access or CPU utilization. I'm even susceptible to it on a Q6700. I've taken measures to counter it by using XFS's agsize as a multiple of my throughput/sec. I know there are some scheduling tweaks posted by kernelOfTruth over in the tuning thread for ReiserFS/Reiser4. Properly formating your disk/partition (cluster size of 4K will have less latency but more overhead than a cluster size of 64K) with cryptcompress can make a sizable difference in performance and realtime usage. I'm still tuning and tweaking tmgr.atom_max_size and tmgr.atom_max_age to see where I can get some stability and fluidity in disk access.


I've just used create=ccreg40,compress=lzo1, no changes to default cluster sizes etc. And no tweaks to any scheduler/fs.

Accessing just source code or mostly source code? If you have a HDD partitioned for special purposes to help with performance and have just your source code on one I'd suggest backing it up and reformatting with the following additional options
Code:
fibration=dot_o_fibre,cluster=8K
This will organize the files better for compiling purposes and allow finer grained access to chunks/clusters of files which should help out a bit on performance
neuron wrote:
I'm using device mapper encryption however, could be they dont mix all that well. I realized ext4+luks seems to have way less latency issues than r4+luks does on my system. With ext4 I can barely notice, with r4 my mouse occasionally stops moving.

Note that these are drives on the same system, but different partitions, so the results are not directly comparable.

Read a couple posts up. kernelOfTruth notes that r4+luks don't mix well. I don't have experience with that so I cannot comment on performance or stability. It would be reason to assume that adding additional encryption, especially software based will hurt performance a lot. I will add two more cents to say that from experience, if you have a crash on a Reiser4 + cryptcompress system, any files in that FS that are being accessed will be orphaned. If they are crucial files for boot or what have you, then hope you have a backup because it is painful to try and sort through them. Adding excryption on top of that could cause some more serious problems.
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://mobrienphotography.com/
Back to top
View user's profile Send private message
kernelOfTruth
Watchman
Watchman


Joined: 20 Dec 2005
Posts: 6111
Location: Vienna, Austria; Germany; hello world :)

PostPosted: Mon Feb 16, 2009 9:37 am    Post subject: Reply with quote

DigitalCorpus wrote:

Read a couple posts up. kernelOfTruth notes that r4+luks don't mix well. I don't have experience with that so I cannot comment on performance or stability. It would be reason to assume that adding additional encryption, especially software based will hurt performance a lot. I will add two more cents to say that from experience, if you have a crash on a Reiser4 + cryptcompress system, any files in that FS that are being accessed will be orphaned. If they are crucial files for boot or what have you, then hope you have a backup because it is painful to try and sort through them. Adding excryption on top of that could cause some more serious problems.


currently I'm experimenting with cfq-settings and reiser4 mount settings, hopefully I'll gain some responsiveness / performance

I'll post my results if they make any change compared to elder ones
_________________
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 :D
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 283

PostPosted: Mon Feb 16, 2009 9:43 am    Post subject: Reply with quote

kernelOfTruth wrote:
DigitalCorpus wrote:

Read a couple posts up. kernelOfTruth notes that r4+luks don't mix well. I don't have experience with that so I cannot comment on performance or stability. It would be reason to assume that adding additional encryption, especially software based will hurt performance a lot. I will add two more cents to say that from experience, if you have a crash on a Reiser4 + cryptcompress system, any files in that FS that are being accessed will be orphaned. If they are crucial files for boot or what have you, then hope you have a backup because it is painful to try and sort through them. Adding excryption on top of that could cause some more serious problems.


currently I'm experimenting with cfq-settings and reiser4 mount settings, hopefully I'll gain some responsiveness / performance

I'll post my results if they make any change compared to elder ones

Cool. I await your posts in the tuning thread
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://mobrienphotography.com/
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Unsupported Software All times are GMT
Goto page Previous  1, 2, 3, 4, 5, 6, 7  Next
Page 2 of 7

 
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