Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
ReiserFS tuning thread, the mother of all "ricer" threads ;)
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3, 4, 5  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 245

PostPosted: Thu Feb 19, 2009 6:14 am    Post subject: Reply with quote

I've found iotop to be fairly reliable. Unfortunately because of my HDDs going upwards of 110-120MB/sec, I don't have a bottleneck on throughput at the disk level. I might have a chance to pop in an old PATA drive that is roughly limited to ~30MB/sec so I hope your results pan out.
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
Back to top
View user's profile Send private message
darklegion
Guru
Guru


Joined: 14 Nov 2004
Posts: 468

PostPosted: Thu Feb 19, 2009 7:59 am    Post subject: Reply with quote

neuron wrote:
kernelOfTruth wrote:
Quote:
for i in /sys/block/sd*; do
/bin/echo "cfq" > $i/queue/scheduler
/bin/echo "0" > $i/queue/iosched/slice_idle
/bin/echo "64" > $i/queue/iosched/quantum
done


and ideally (if you're having drives with ncq)
Code:
for i in /sys/block/sd*; do
         /bin/echo "1024" >  $i/queue/nr_requests
         /bin/echo "2" >  $i/device/queue_depth
done



I get the best results with them so far ...


I only have read only access to queue_depth, has that changed in recent kernels or something like that? It's 1, and wont change to 2.


You probably need to enable AHCI in your bios.I had to with my GA-ep45-ds3; I have to wait an extra 5+ seconds for an annoying Intel AHCI Bios to load, though.The hardware side of that board is fantastic, but damn does the bios suck!
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 245

PostPosted: Sat Feb 28, 2009 7:54 am    Post subject: Reply with quote

Third time I'm almost set with what I have and then something has to go wrong and lock the system. fsck keeps seg faulting so my /root is toasted in odd locations. Seems though I still have a good /etc and a portage world file so I can rebuild without much of anything but time. From something that kerneloftruth noted earlier about reiser4+ cryptcompress, I'm going to give it a shot without compression and see if I have a more stable system, at least if something goes wrong I'm not fsck'd.

I'm really ticked because the literal next thing I was going to do was make a backup... :evil:

I want to start making benchmarks and doing some tests to help out R4 to avoid this
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
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: Sat Feb 28, 2009 11:14 am    Post subject: Reply with quote

hm, sorry to hear that

seems like you're running kind of an "unstable" reiser4 kernel (unstable for reiser4)

there are kernel-releases or kernel-patches from time to time that leave a somewhat unstable and often error-throwing system

for me this is often because I'm running custom CFLAGS (which I stopped doing) because its not worth the extra gain of speed, security, ...


this shows again that reiser4 really needs to get into mainline that it gets more testing in a short period of time and lots of outstanding (hidden) BUGs fixed


if you're running vanilla 2.6.28* + reiser4-patch, please next time try 2.6.28-zen10 for me it's running rocksolid (give it a try - if it's more unstable ditch it)

I still have to try copying over big amounts of files / creating a stage4-tarball from my system on an cryptcompressed partition to be really sure though ...

before that I'll create a backup reiser4+ccreg -> reiserfs
_________________
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: 245

PostPosted: Sat Feb 28, 2009 8:35 pm    Post subject: Reply with quote

Well I've been consistent with using 2.6.28 Gentoo Sources patched with Reiser4 through the ebuild provided here on these forums support for about 4 months. My CFLAGS have been consistently[code]-march=native -O2 -pipe -mno-push-args -mcx16 -msahf[code] though I've jumpped between that and core2 for march a few times. The 'instability' isn't the fault of Reiser4, but other pieces of software that for example have not loaded correctly when openrc engages like what I had with udev recently. That caused me to do resets to the system which is probably when it got trashed. Because of reiser4 + cryptcompress disconnecting accessed files from the fs structure until a flush any minor problems can cause others of an entirely different nature due to the corruption I guess.

As for actual application related stability and not configuration issues, I've only had one problem and that boiled down to my overclock needeing just a smidgen more voltage for Vcore.

After backing everything up I was going to test out the latest zen sources for the new scheduler BFQ.
[/code]
I've never had a problem with R4 + CC with my portage partition so I'll still be using it immensely on that end. Plus that data is easily fixable ;).
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
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 Mar 02, 2009 9:37 pm    Post subject: Reply with quote

anyone tried tweaking the cbk-cache ?

Quote:
/* &cbk_cache - coord cache. This is part of reiser4_tree.

cbk_cache is supposed to speed up tree lookups by caching results of recent
successful lookups (we don't cache negative results as dentry cache
does). Cache consists of relatively small number of entries kept in a LRU
order. Each entry (&cbk_cache_slot) contains a pointer to znode, from
which we can obtain a range of keys that covered by this znode. Before
embarking into real tree traversal we scan cbk_cache slot by slot and for
each slot check whether key we are looking for is between minimal and
maximal keys for node pointed to by this slot. If no match is found, real
tree traversal is performed and if result is successful, appropriate entry
is inserted into cache, possibly pulling least recently used entry out of
it.

Tree spin lock is used to protect coord cache. If contention for this
lock proves to be too high, more finer grained locking can be added.

Invariants involving parts of this data-type:

[cbk-cache-invariant]
*/



Quote:
Mount options
=============

[...]
tree.cbk_cache.nr_slots=N
Number of slots in the cbk cache.


default in decimal number seems to be 16 (0x10)
_________________
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: 245

PostPosted: Tue Mar 03, 2009 5:43 am    Post subject: Reply with quote

to benchmark that you'd have to read random files once and then again to see it's affect it seems. I did some testing when I was trying to tweak performance and syncing/flushing to disk and didn't notice anything in particular. Maybe edward can point us to how to test it?
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
Back to top
View user's profile Send private message
lysergicacid
Guru
Guru


Joined: 25 Nov 2003
Posts: 352
Location: The Universe,Virgo Super Cluster,Milky Way,Earth

PostPosted: Wed Mar 11, 2009 5:40 am    Post subject: Reply with quote

any ideas why this is happening please ?
Code:
UniMatrix0 linux # dd if=/dev/zero of=/mnt/swap.img count=512 bs=1M
512+0 records in
512+0 records out
536870912 bytes (537 MB) copied, 4.96409 s, 108 MB/s
UniMatrix0 linux # mkswap -v1 /mnt/swap.img
Setting up swapspace version 1, size = 524284 KiB
no label, UUID=499ff382-fd20-43bc-8a11-49413a012d94
UniMatrix0 linux # swapon /mnt/swap.img
Adding 524280k swap on /mnt/swap.img.  Priority:-1 extents:131071 across:4k


works until it wants use it:

Code:
Write-error on swap-device (3:2:18446744073709551440)
attempt to access beyond end of device
hda2: rw=1, want=18446744073709551448, limit=20000925
Write-error on swap-device (3:2:18446744073709551440)
attempt to access beyond end of device
hda2: rw=1, want=18446744073709551448, limit=20000925
Write-error on swap-device (3:2:18446744073709551440)


this is with these reiser4 options:
Code:
mkfs.reiser4 -o create=ccreg40,compress=lzo1,compressMode=force,cluster=8K,fibration=lexic_fibre,formatting=smart /dev/hda2

_________________
[img]http://valid.canardpc.com/cache/banner/2040927.png[/img]
Desktop:
[img]http://valid.canardpc.com/cache/banner/2703952.png[/img]
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 Mar 11, 2009 10:51 am    Post subject: Reply with quote

a recommendation: you shouldn't use compressMode=force anymore since it might bloat files and slow down processing of files, like energyman write earlier, I've also made this experience

so

mkfs.reiser4 -o create=ccreg40,compress=lzo1,compressMode=force,cluster=8K,fibration=lexic_fibre,formatting=smart /dev/hda2

should become

mkfs.reiser4 -o create=ccreg40,compress=lzo1,compressMode=ultim,cluster=8K,fibration=lexic_fibre,formatting=smart /dev/hda2

mkfs.reiser4 -o create=ccreg40,compress=lzo1,compressMode=conv,cluster=8K,fibration=lexic_fibre,formatting=smart /dev/hda2

or

mkfs.reiser4 -o create=ccreg40,compress=lzo1,compressMode=latt,cluster=8K,fibration=lexic_fibre,formatting=smart /dev/hda2

(I have no experience with lattice algorithm)

I'm experimenting with cluster=64K (default) and some other settings right now


perhaps you're running into that problems because reiser4 doesn't support swapfile or because of the compression

anyways, please try without compressMode=force

I've you're still running into that problem report it to edward / the reiserfs mailing list

what kernel-version are you running ?

2.6.28* or 2.6.29-rc*-zen or mmotm is recommended to get latest reiser4 version with fixes
_________________
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


Last edited by kernelOfTruth on Wed Mar 11, 2009 10:56 am; edited 1 time in total
Back to top
View user's profile Send private message
neuron
Advocate
Advocate


Joined: 28 May 2002
Posts: 2371

PostPosted: Wed Mar 11, 2009 10:54 am    Post subject: Reply with quote

lysergicacid wrote:
any ideas why this is happening please ?


You cant use compressed loopback images.
Back to top
View user's profile Send private message
lysergicacid
Guru
Guru


Joined: 25 Nov 2003
Posts: 352
Location: The Universe,Virgo Super Cluster,Milky Way,Earth

PostPosted: Wed Mar 11, 2009 10:16 pm    Post subject: Reply with quote

ok thanks guys :) have formated another partiction without the force option as sugested and i guess i'll have reparticion for a swap particion. the kernel sources are
Code:
hitchhiker-r3

_________________
[img]http://valid.canardpc.com/cache/banner/2040927.png[/img]
Desktop:
[img]http://valid.canardpc.com/cache/banner/2703952.png[/img]
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: Fri Mar 20, 2009 2:30 pm    Post subject: Reply with quote

Is there an official (or semi-official) RTFM site?

This forum post is about as close as I've seen.

Reading through the options for making a reiser4 fs discussed in this thread (and looking through the man page), I've become confused about something.

Is compression the default for reiser4 (i.e. "mkfs.reiser4 /dev/sdx1" with no other options)?

What would be the syntax if I wanted to explicitly create a reiser4 fs without compression?
Back to top
View user's profile Send private message
jordanwb
l33t
l33t


Joined: 10 Jul 2008
Posts: 642
Location: Ottawa, Canada

PostPosted: Fri Mar 20, 2009 2:38 pm    Post subject: Reply with quote

Dont Panic wrote:
What would be the syntax if I wanted to explicitly create a reiser4 fs without compression?


Good question.
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 Mar 20, 2009 4:14 pm    Post subject: Reply with quote

jordanwb wrote:
Dont Panic wrote:
What would be the syntax if I wanted to explicitly create a reiser4 fs without compression?


Good question.


creating a reiser4-partition with the default settings creates the "former" or default reiser4 kind of filesystem without compression:

in the simplest manifestation:

Code:
mkfs.reiser4


(this would call mkfs.reiser4 with the command -o create=reg40 which is default)

Quote:
"reg40", "unix-file plugin". Manages regular files.


unix-file plugin provides reiser 4.0 filesystem functionality (no compression but some neat features like fibration, tail-packing, etc.)



Quote:
"ccreg40", "cryptcompress file plugin". Manages regular files with encrypted and(or) compressed bodies.


should equal the formally featured reiser 4.1 filecompression featureset (of course there was more like xattr but that will follow soon from what I've read)


o.k., let me take an example:

Code:
mkfs.reiser4 -o create=ccreg40,compress=gzip1,compressMode=ultim,cluster=8K,fibration=lexic_fibre,formatting=smart


would create an reiser4 partition with the ccreg40 (cryptcompress) plugin with gzip-compression, ultim compression check-algorith, a cluster-size of 8 KB (default 64K), lexicographic file ordering and an automatic algorithm which decides whether it converts stuff to a tail (saving space but causing a little performance degradation from 1-5 %) or to extents

now you want a pretty same filesystem but only without the compression

Code:
mkfs.reiser4 -o cluster=8K,fibration=lexic_fibre,formatting=smart


so you're leaving out
Quote:
create=ccreg40,compress=gzip1,compressMode=ultim,


so the following options define the cryptcompress-plugin / -handling:

- create=ccreg40
- compress=
- compressMode=



there's also a way to format the partition with the cryptcompress-plugin but without compression

compressMode=none

I don't know how that differs from the unix-file plugin besides additional checksumming


there's also a way in between

compressMode=conv (default in cryptcompress) so if a file isn't seen as fit to be handled by the compression algorithm it is passed to the unix-file plugin (extents way of handling),


now there's again another derivative from that

if you called formatting=tails instead of formatting=smart in that way, I guess, stuff gets converted to a tail instead of an extent



DISCLAIMER:

I haven't read much of the code or only very little of it so that's mostly interpretation and to clear things up

if you need an in-depth view

*) RTFC (read the fine code) ;)
*) ask Edward
_________________
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
kernelOfTruth
Watchman
Watchman


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

PostPosted: Sat Apr 04, 2009 8:52 pm    Post subject: Reply with quote

what ? still not satisfied :P



....

anyways here are some settings I'm currently using (and experimenting) which I got from the amd64 subforum (those folks having latency / responsiveness problems under heavy i/o might already know them):

no definite comments yet since I'm still pretty ambivalent about them (meaning I'm getting no definite results):

I use the following in /etc/conf.d/local.start:

Code:
for i in `pidof kjournald` ; do ionice -c1 -p $i ; done
for i in `pidof kjournald2` ; do ionice -c1 -p $i ; done
for i in `pidof kcryptd` ; do ionice -c1 -p $i ; done
for i in `pidof ktxnmgrd` ; do ionice -c1 -p $i ; done
for i in `pidof ent` ; do ionice -c1 -p $i ; done
for i in `pidof pdflush` ; do ionice -c1 -p $i ; done


(ktxnmgrd, ent are specific to reiser4, pdflush should improve reiser4' heavy flushing somewhat)


you can also try to renice pdflush and kcryptd for the cpu scheduler via:

Code:
for i in `pidof pdflush` ; do renice -20 $i ; done
for i in `pidof kcryptd` ; do renice -20 $i ; done



make sure you disable hald-addon-storage polling [e.g. killall -9 hald-addon-storage] (in some cases this can improve throughput)


best reiser4-partition creating options are so far:

mkfs.reiser4 -o create=ccreg40,compress=gzip1,compressMode=ultim,cluster=8K,fibration=lexic_fibre,formatting=smart for /usr/portage

and

mkfs.reiser4 -o create=ccreg40,compress=gzip1,compressMode=ultim,cluster=8K,formatting=smart (without the lexic ordering) for / (root / or system how you like to call it)


mount-options are so far:

noatime,nodiratime,tree.cbk_cache.nr_slots=48 (you might need to lower tree.cbk_cache.nr_slots to 24 or 32 if you have low memory, dunno right now how memory usage depends on it)
_________________
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
dreadlorde
Apprentice
Apprentice


Joined: 16 Dec 2008
Posts: 243
Location: /adm/timezone/US_Michigan

PostPosted: Sun Apr 05, 2009 3:00 pm    Post subject: Reply with quote

kernelOfTruth wrote:


cluster=8K,fibration=lexic_fibre,formatting=smart for /usr/portage

What do these options do?

And which type of compression in reiser4 compresses data more? lzo1 or gzip1?

This is how much I got with lzo1 on my /usr/portage partition
Code:

Total Disk Usage: 444.4 Mb
Apparent Size: 394.1 Mb
Items: 158345

vs. no compression
Code:

Total Disk Usage: 894.5 Mb
Apparent Size: 513.5 Mb
Items: 158345

_________________
Ludwig von Mises Institute
Quote:
I am not to be a shepherd, I am not to be a grave-digger. No longer will I speak to the people; for the last time I have spoken to the dead.
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: Sun Apr 05, 2009 3:11 pm    Post subject: Reply with quote

dreadlorde wrote:
kernelOfTruth wrote:


cluster=8K,fibration=lexic_fibre,formatting=smart for /usr/portage

What do these options do?




please have a look what DigitalCorpus wrote (on first page):

https://forums.gentoo.org/viewtopic-p-5212962.html#5212962


dreadlorde wrote:


And which type of compression in reiser4 compresses data more? lzo1 or gzip1?

This is how much I got with lzo1 on my /usr/portage partition
Code:

Total Disk Usage: 444.4 Mb
Apparent Size: 394.1 Mb
Items: 158345

vs. no compression
Code:

Total Disk Usage: 894.5 Mb
Apparent Size: 513.5 Mb
Items: 158345


of course gzip compresses more, lzo is more kind of light-weight realtime compression and decompression, gzip is more kind of archive compression (for backup or simply data storage purposes)
_________________
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
Strowi
l33t
l33t


Joined: 19 Aug 2003
Posts: 655
Location: Bonn

PostPosted: Thu May 21, 2009 1:00 am    Post subject: Reply with quote

[quote="kernelOfTruth"]
best reiser4-partition creating options are so far:

[b]mkfs.reiser4 -o create=ccreg40,compress=gzip1,compressMode=ultim,cluster=8K,fibration=lexic_fibre,formatting=smart[/b] for /usr/portage

and

[b]mkfs.reiser4 -o create=ccreg40,compress=gzip1,compressMode=ultim,cluster=8K,formatting=smart[/b] (without the lexic ordering) for / (root / or system how you like to call it)


mount-options are so far:

[b]noatime,nodiratime,tree.cbk_cache.nr_slots=48[/b] (you might need to lower tree.cbk_cache.nr_slots to 24 or 32 if you have low memory, dunno right now how memory usage depends on it)[/quote]
I am running 2.6.29-zen2 with 'default' reiser4+lzo1 quiete stable for a long time.
Since i had some time, i tested those settings, and everytime i get fs-corruptions after a couple of reboots...
_________________
--
Linux & such ...
http://blog.hasnoname.de
Back to top
View user's profile Send private message
Colt45
Tux's lil' helper
Tux's lil' helper


Joined: 05 Sep 2007
Posts: 102
Location: Central Washington

PostPosted: Tue Jul 21, 2009 11:24 pm    Post subject: Reply with quote

I just had an interesting experiance with r4. I had my /home partition formatted with "lzo1" compression, "ultim" compress mode, "16K" cluster, and "smart" extents. The inital problem was with the ssh known_hosts file. It existed, but it didnt. It showed up in the folder but was blinking red, 0 byte and impossible to delete. I ran a fsck and it told me there were errors and to run --build-fs, which I did. Oops. It flooded the screen with errors and it went downhill from there. kde wouldnt start, many folders gave I/O errors if I tried to look in them. I ended up just reformatting it. But I kinda expected it and I had much of my folder copied on another partition. All I lost were some game trailers.
I did not use ccreg40 this time. :P The rest of the drive is compressed, however.
_________________
"Hopps" - AMD Athlon X2 4450B, 4GB, 128GB Samsung SSD, (Gentoo Hardened amd64; gcc-4.9.3; 4.4.8-hardened-r1)
"Dash" - 2x Intel Xeon 5148, 12GB FB-DIMM, 3x Samsung 1TB (mdadm RAID 5), AMD HD6950, (Gentoo amd64-multilib; gcc-5.3.0; 4.4.0-pf4)
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 245

PostPosted: Fri Jul 31, 2009 12:55 pm    Post subject: Reply with quote

You'll have data loss like that if you don't shutdown properly. Try adding
Code:
tmgr.atom_max_age=seconds
to your mount options in /etc/fstab to help items get flushed to disk sooner to avoid this. I use 60 on my /root partition. Also if you have a file that is blinking red, I believe that is a broken symlink, that is one file point to another but that other is not existant. So it sounds like you had more than one thing wrong and that compounded with Reiser4. Reiser4 may not have been at fault. I'm running 129 days of total uptime (42 consecutive) with only 3 down.

With adding the aforementioned mount option, any time I've had a crash or power outage, I'm sustained no data loss on these partitions. Also keep in mind that other file systems like Ext4 suffer from the same cause of data loss that Reiser4 does with or without cryptcompress being used.
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
Back to top
View user's profile Send private message
seqizz
Tux's lil' helper
Tux's lil' helper


Joined: 14 Jan 2008
Posts: 103

PostPosted: Fri Oct 02, 2009 11:55 am    Post subject: Reply with quote

i don't know why this is happening, but since i've started to using reiser4 for my root folder (/) , i'm taking some lockups when using resources heavily. Also i made a change on fs (e.g edit a file) and after one reboot, it's turning back to old one, like i've never touched the file.. (wtf? :mrgreen: ) i think i should change my mount options, btw i'm using KernelOfTruth's last recipe..
_________________
what about an answer here?
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 Oct 02, 2009 12:20 pm    Post subject: Reply with quote

so for those of you who have problems with files disappearing (I know this kind of "feature" - I'm also experiencing it with ext4 and xfs :mrgreen: ):

noatime,nodiratime,tree.cbk_cache.nr_slots=48,tmgr.atom_max_age=60

the default is 600 (10 minutes)

I'm manually syncing from time to time to ensure important stuff is on the harddrive so perhaps that helps :wink:

also I'm rebooting my system the following way:

Code:
su
time sync && reboot


that way I see how long it took to write data and that really stuff was commited to disk before reboot

to check whether there's data that needs to be written to disk

Code:
cat /proc/meminfo


Quote:
Dirty: 556268 kB
Writeback: 872 kB

<-- this show that there's 556000 kB to be written and currently (more or less) is writing 872 kB to the disk


a WARNING for those of you who are using reiser4 with zen-sources

it's currently broken with 2.6.31-zen1 and 2.6.31-zen2 and will be fixed soon so you better backup your data :idea:

this happens from time to time since upstream is changing stuff in the kernel or other patches / components in zen are interacting with reiser4

this e.g. leads to the following behavior:

trying to emerge openoffice with /var/tmp/portage on an partition with reiser4 leads to D (daemonized) tar which doesn't continue extracting a tar-bz2 file during extraction phase ...

@Strowi:

with reiser4 and zen always the following is recommended:
try to run latest stable release (right now this should be around 2.6.30-zen*) whether that fixes it

you also might be suffering from a kernel (non-reiser4 specific) or hardware issue ...

good luck guys :)
_________________
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
unK
l33t
l33t


Joined: 06 Feb 2007
Posts: 769

PostPosted: Fri Oct 02, 2009 3:58 pm    Post subject: Reply with quote

seqizz wrote:
i don't know why this is happening, but since i've started to using reiser4 for my root folder (/) , i'm taking some lockups when using resources heavily. Also i made a change on fs (e.g edit a file) and after one reboot, it's turning back to old one, like i've never touched the file.. (wtf? :mrgreen: ) i think i should change my mount options, btw i'm using KernelOfTruth's last recipe..

what kernel do you use? I once had such behaviour with zen-sources.
_________________
ncmpcpp - featureful ncurses based MPD client inspired by ncmpc
Back to top
View user's profile Send private message
DigitalCorpus
Apprentice
Apprentice


Joined: 30 Jul 2007
Posts: 245

PostPosted: Fri Oct 02, 2009 5:30 pm    Post subject: Reply with quote

KernelOfTruth, can you elaborate on how R4 is broken in Zen sources currently?
_________________
Atlas (HDTV PVR, HTTP & Media server)
http://atlas.selfip.net/Info/
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 Oct 02, 2009 6:20 pm    Post subject: Reply with quote

DigitalCorpus wrote:
KernelOfTruth, can you elaborate on how R4 is broken in Zen sources currently?


I can't say much about its current state:

*) given the nature of reiser4 and data integrity in mind when it was designed your data should be safe on reiser4-partitions (no guarantee)

*) reiser4 with cryptcompress and gzip was more or less fine on / (root) but it showed some slow-down in operations and wouldn't really sync data when invocing the "sync" command unless the system was shutdown or rebooted (this was only partially, most of the time "sync" of data was fine)

*) when extracting or transferring large amounts of data (e.g. from a stage4-tarball or via rsync) after some amount (some gigabytes) it would just stall there and not continue (resulting in cpu-load of 0%)
-> so the problem is / was that data couldn't be copied over to a reiser4-partition since it wouldn't "want" to store it

*) trying to extract a portage-tarball (mine is around 500 MB big in tarbz2 archive) it wouldn't extract it completely to the /usr/portage partition

*) in the above mentioned case of trying to compile openoffice it wouldn't even reach the compile phase because the tar-command would show D (daemon / not killable) on the package sys-libs-extern (or what its name was) and thus simply stalling the process leading to not extraction of the mentioned tarball and not continuing
-> killing all of the other processes (ebuild.sh, make, etc.) and then a forceful umount would be successful but the endresult is unsatisfying: you can't really work with reiser4-partitions or (large amounts of) data
-> this was applicable to the system-partition and separate test-partitions for /var/tmp/portage or other data-partitions: I tried to emerge it with /var/tmp/portage on the system-partition and on an separate external partition


Conclusion:
=> reiser4 finally needs to get into mainline that it can be fixed ASAP as breakage occurs like with the other filesystems
-> it's doing a great job in preserving data integrity
_________________
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
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  Next
Page 3 of 5

 
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