___DEPRECATED___ 2.6.14-nitro2

Message

Tiger683 · Post by **Tiger683** » Fri Nov 18, 2005 7:49 pm

i didn't use cfq since ages

Thetargos · Post by **Thetargos** » Fri Nov 18, 2005 7:55 pm

But CK5 puts CFQ as the default sched

Thetargos · Post by **Thetargos** » Fri Nov 18, 2005 8:20 pm

This may be unrelated, but I thought I'd ask... I was looking at the help item for the CONFIG_LOG_BUF_SHIFT item in the Kernel Hacking section of menuconfig, and was wondering what would the best setting be... Te help says that for uniprocessor systems (i386, pressumably) the best option is to set it to 14 (a 16 Kb size) and for SMP 15 (32 Kb size)... However I'm inrigured since I'm configuring my kernel and others for x86_64 systems, so naturally I'd like to know which is best for this architecture, I see the 64 Kb size (number 16) for IA64 and x86 NUMAQ, should I use this for x86_64 too? And what would this do, anyway?

TIA!

yaneurabeya · Post by **yaneurabeya** » Sat Nov 19, 2005 4:28 am

Thetargos wrote:Bur you are using ReiserFS 3.6, don't you? Because Reiser4 requires you to have ReiserProgrs 4

Yes. I wouldn't touch reiser4 with a 10 ft pole because it's such a shoddy fs type.

iphitus · Post by **iphitus** » Sat Nov 19, 2005 7:05 am

yaneurabeya wrote:
Thetargos wrote:Bur you are using ReiserFS 3.6, don't you? Because Reiser4 requires you to have ReiserProgrs 4
Yes. I wouldn't touch reiser4 with a 10 ft pole because it's such a shoddy fs type.

It's quite understandable that Reiser4 requires reiserprogs 4. It's an entirely new filesystem, not a line of important code was copied from Reiser3. It was written from scratch.

As for calling it shoddy, that's a pretty weak and lame attack at something you know little about.

You realise that every filesystem went through a stage where it was problematic and/or unstable?

iphitus

HecHacker1 · Post by **HecHacker1** » Sat Nov 19, 2005 7:07 am

yaneurabeya wrote:
Thetargos wrote:Bur you are using ReiserFS 3.6, don't you? Because Reiser4 requires you to have ReiserProgrs 4
Yes. I wouldn't touch reiser4 with a 10 ft pole because it's such a shoddy fs type.

the problems I was having with reiser4 is unrelated to this kernel and unrelated to the cfq schedule, I too use anticipatory. I have been using reiser4 for a long time without any data loss. When it works right, it does surpass the other filesystems in terms of performance, and its a tangible performance gain, not just numbers.

I think it's just a problem when creating a new reiser4 partition using the 1.0.5 tools. Most operations on a 1.0.5. partition are ok, but some things like flushing to disk take forever, 30+ seconds compared to ~1 sec. If you create the partition using 1.0.0 tools then performance is top notch, but I'm not sure about compatibility between the 1.0.5 tools and the 1.0.0 created partitions, but appears that many of you are having no problems running them mixed.

Tiger683 · Post by **Tiger683** » Sat Nov 19, 2005 10:47 am

I have created my partitions with 1.0.5 too, and , well,
i can't reproduce any issues you guys are having...

Could it be that you converted from previous reiser4 version?
because i didn't.....

T

1U · Post by 1U » Sat Nov 19, 2005 4:47 pm

yaneurabeya wrote:
Thetargos wrote:Bur you are using ReiserFS 3.6, don't you? Because Reiser4 requires you to have ReiserProgrs 4
Yes. I wouldn't touch reiser4 with a 10 ft pole because it's such a shoddy fs type.

Reiser4 is an awesome fs, and I use it everywhere without ANY problems (thanks to Tiger and nitro-sources). Please don't post your opinionated bs which you gained from reading rantings by neophobic developers. Hans Reiser is doing a great job.

yaneurabeya · Post by **yaneurabeya** » Sat Nov 19, 2005 5:03 pm

I realize that all things take a particular amount of time in development cycles to come to maturity; that's the nature of being in terms of software.

What I meant by 'shoddy' is that I wouldn't use it on a production system because I've seen lots of people on this forum lose data to Reiser4 because they used it on their systems, since it's still in the developmental/unstable stages. Besides, if it was safe to use it would be included in the standard kernel sources available on kernel.org.

That above provides the basis for my opinion, and I didn't mean any attack on anyone's grounds and I'm going to end my part of the discussion now. If it works for you, wonderful, but I just don't want to risk losing data because I've already lost enough data to convential methods and other fs'es already.

1U · Post by 1U » Sat Nov 19, 2005 5:13 pm

Losing data like you said can happen with any fs, if you didn't have backups then that's your fault. Reiser4 is as safe as other fses at this moment. Yes it was very fresh at one point, but don't expect the kernel developer's judgement to be correct on when it should and should not be included in the kernel. They have personal problems with Hans Reiser so I'm pretty sure Microsoft will support Reiser4 before than vanilla sources will.

Jedi Master · Post by **Jedi Master** » Sat Nov 19, 2005 5:48 pm

In my experience with ext3 I never lost any data except 1 case - when HD crashed physically with about 1GB of bad blocks. Even working on laptop, battery powered - up to the last bit of the battery, when system was powered down in the middle of disk write activity. The system is tested and found rock stable. I had 2 experiments with XFS and Reiser4 - in both cases battery down condition eneded up in COMPLETE data loss. Not just file/directory converted to lost+found - file system just crashed, dead, no more, not even a tiny file recovered. So I do not play reiser any more - if you have server with decent UPS and can be shure your box will shutdown normally, and need fs speed most of all, and have a mirror and full backup daily - thats probably a wise choice. Not for me. I'll better wait 1 extra minute at bootup rather than spend a whole day restoring the backup.

1U · Post by 1U » Sat Nov 19, 2005 6:46 pm

Not true. I have improperly shut down machines like you at least 20 times without resulting in any file loss on Reiser4 partitions.

What do you mean by having 2 "experiments" with Reiser4 and xfs? Your only goal was to crash them? I don't experiment, I UTILIZE... which usually ends up with good results due to my decision process which does not involve reading opinionated complaints and being extra cautious at the expence of not moving ahead.

Jedi Master · Post by **Jedi Master** » Sat Nov 19, 2005 7:07 pm

I just ment to try an alternative - and start UTILIZING. And end up with restoring from backup. With ext3 I am using for 2 years - no crash except the one I mentioned. With reizer and xfs it was a week or so. In my case - laptop battery no power the inproper shutdown is dayly - not 20 cases.

Tiger683 · Post by **Tiger683** » Sat Nov 19, 2005 7:59 pm

using ext3 with journal writeback has about 10 times higher risk of
having old data appear in the modified AND saved files after a powerloss
than with reiser4 .... Considering the fact, reiser4 also uses writeback, let's see which one has
worse implementation

Then , even your wonderful dir_index won't help...

Oh, and did you even consider mounting reiser4 with setting the intervall
for flushing data to disk as often as ext3? because you can

iphitus · Post by **iphitus** » Sat Nov 19, 2005 10:42 pm

My primary computer is a laptop, and I used reiser4 for about six months with no problems. And I can tell you that I lost count of how many times my battery ran dead without me shutting down properly. R4 was solid.

I'm using a mix of R3 and ext3 right now, and I havnt had any issues either.

b3cks · Post by **b3cks** » Sat Nov 19, 2005 10:45 pm

iphitus wrote:My primary computer is a laptop, and I used reiser4 for about six months with no problems. And I can tell you that I lost count of how many times my battery ran dead without me shutting down properly. R4 was solid.

I'm using a mix of R3 and ext3 right now, and I havnt had any issues either.

Agreed.

mayday147 · Post by **mayday147** » Sun Nov 20, 2005 1:15 am

Jedi Master wrote:I just ment to try an alternative - and start UTILIZING. And end up with restoring from backup. With ext3 I am using for 2 years - no crash except the one I mentioned. With reizer and xfs it was a week or so. In my case - laptop battery no power the inproper shutdown is dayly - not 20 cases.

I'm using reiser4 and XFS on both my system (one desktop and the other one a laptop) and NEVER had any problems.

P.S. There goes a FS flamewar starting on

1U · Post by 1U » Sun Nov 20, 2005 2:13 am

mayday147 wrote:P.S. There goes a FS flamewar starting on

I've already come and done. Join in.

Btw: Can someone help me with this tulip problem?
http://forums.gentoo.org/viewtopic-t-40 ... ight-.html
I may or may not be the nitro-sources patchset's fault but that's what I run and nobody seems to know enough about the problem I posted to reply. It's really beginning to piss me from the occasional 20 minutes of internet downtime every 1-24 hours.

HecHacker1 · Post by **HecHacker1** » Sun Nov 20, 2005 7:29 am

Tiger683 wrote:I have created my partitions with 1.0.5 too, and , well,
i can't reproduce any issues you guys are having...

Could it be that you converted from previous reiser4 version?
because i didn't.....

T

Originally I did try converting from 1.0.0 to 1.0.5 but that procedure made my reiser4 partition crash, it worked fine for a couple weeks, but then all of the sudden exploded.

That is when I tried starting with the 1.0.5 tools to make a new paritition. That partition was reliable as far as data is concerned, however too slow. Maybe after a few reiser4 patches and updates i'll go back to it.

bollucks · Post by **bollucks** » Mon Nov 21, 2005 1:46 am

Tiger683 wrote:using ext3 with journal writeback has about 10 times higher risk of
having old data appear in the modified AND saved files after a powerloss
than with reiser4 .... Considering the fact, reiser4 also uses writeback, let's see which one has
worse implementation
Then , even your wonderful dir_index won't help...

Where did you get your "information" from saying the data loss will be less?
Sounds like bullshit handwaving to me.

1U · Post by 1U » Mon Nov 21, 2005 1:48 am

bollucks wrote:Where did you get your "information" from saying the data loss will be less?
Sounds like bullshit handwaving to me.

He actually made a statement, if you disagree then back up your point of view instead of posting mind numbing garbage.

bollucks · Post by **bollucks** » Mon Nov 21, 2005 1:53 am

1U wrote:
bollucks wrote:Where did you get your "information" from saying the data loss will be less?
Sounds like bullshit handwaving to me.
He actually made a statement, if you disagree then back up your point of view instead of posting mind numbing garbage.

He said clearly: using ext3 with journal writeback has about 10 times higher risk of
having old data appear in the modified AND saved files after a powerloss
than with reiser4

That sounds like a claim to me. I made no claim.

1U · Post by 1U » Mon Nov 21, 2005 2:14 am

Reiser4 is an atomic filesystem, you can read more about it on the namesys website. I can certainly verify that Reiser4 does not loose data at least as much as all the neophobic anti Reiser4 people claim since I've improperly rebooted my system about 20 times by now on this machine alone.

bollucks · Post by **bollucks** » Mon Nov 21, 2005 2:45 am

1U wrote:Reiser4 is an atomic filesystem, you can read more about it on the namesys website. I can certainly verify that Reiser4 does not loose data at least as much as all the neophobic anti Reiser4 people claim since I've improperly rebooted my system about 20 times by now on this machine alone.

Ok thanks.

Fortunately I've shut down ext3 inappropriately hundreds of times with dodgy kernels, and run it for years on many hard drives as well without any loss of data ever so that is not a concern of mine. I can't say that about my reiser3 experience though

edit: save me the trouble of telling me reiser4 isn't reiser3 - well duh. I never said I judged one by the other, but if a stable filesystem from the same source can't protect data it doesn't bode well for the unstable one (although I still make no hard connection between the two or any claim or yada yada blah blah...)

enderandrew · Post by **enderandrew** » Mon Nov 21, 2005 4:09 am

ReiserFS (or Reiser 3 if you will) did not write atomically. Ext3 and Reiser4 do. Reiser4 should be judged on the merits of Reiser4, not on the merits of ReiserFS.