| View previous topic :: View next topic |
| Author |
Message |
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
 Posted: Wed Apr 06, 2005 3:45 am Post subject: |
 |
|
| venkat wrote: | these are ofcourse my perceptions, ymmv  |
well, they sound reasonable  (unlike some others, and its very well possible these include mine, now and then ) |
|
| Back to top |
|
 |
cokey
Advocate
Joined: 23 Apr 2004
Posts: 3355
|
| Posted: Wed Apr 06, 2005 10:47 am Post subject: |
|
|
| ciaranm wrote: | | feld wrote: | | it WAS stated quite bluntly that Reiser4 WILL SHOW WEAKNESS in hardware --- many drives will break down and die if they cannot handle what it requires due to flaws in the device. |
Ah ha h ah aha ha ha ha ha ha ha ha ha ha ha. |
/me agrees with ciaranm
_________________
https://otw20.com/ OTW20 The new place for off the wall chat |
|
| Back to top |
|
|
MaxDamage
l33t
Joined: 03 Jan 2004
Posts: 650
Location: Oviedo, Spain
|
| Posted: Wed Apr 06, 2005 1:47 pm Post subject: |
|
|
| ciaranm wrote: | | feld wrote: | | it WAS stated quite bluntly that Reiser4 WILL SHOW WEAKNESS in hardware --- many drives will break down and die if they cannot handle what it requires due to flaws in the device. |
Ah ha h ah aha ha ha ha ha ha ha ha ha ha ha. |
Come on, ciaranm, the boy has lost all the data. Don't be bad with him. But now I begin to take seriously your opinion about Reiser4. I'm using ReiserFS 3.6 at home BTW
_________________
La PDA de tungsteno |
|
| Back to top |
|
|
cokey
Advocate
Joined: 23 Apr 2004
Posts: 3355
|
| Posted: Wed Apr 06, 2005 2:06 pm Post subject: |
|
|
| MaxDamage wrote: | | ciaranm wrote: | | feld wrote: | | it WAS stated quite bluntly that Reiser4 WILL SHOW WEAKNESS in hardware --- many drives will break down and die if they cannot handle what it requires due to flaws in the device. |
Ah ha h ah aha ha ha ha ha ha ha ha ha ha ha. |
Come on, ciaranm, the boy has lost all the data. Don't be bad with him. But now I begin to take seriously your opinion about Reiser4. I'm using ReiserFS 3.6 at home BTW |
I think he was laughing more about the comment that reiser4 will show weaknesses in drives any more than any other filesystem. Like an "oogidy boogidy, reiser4 is so powerful it will break your HDD" which is laughable 
_________________
https://otw20.com/ OTW20 The new place for off the wall chat |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Wed Apr 06, 2005 2:09 pm Post subject: |
|
|
| cokehabit wrote: | | I think he was laughing more about the comment that reiser4 will show weaknesses in drives any more than any other filesystem. Like an "oogidy boogidy, reiser4 is so powerful it will break your HDD" which is laughable |
Mostly I was laughing at the amazing levels of outright wrongness that various Hans Ricer fanboys come up with... It's not your hardware, it's the crappy fs. |
|
| Back to top |
|
|
firephoto
Veteran
Joined: 29 Oct 2003
Posts: 1612
Location: +48° 5' 23.40", -119° 48' 30.00"
|
| Posted: Wed Apr 06, 2005 4:04 pm Post subject: |
|
|
| ciaranm wrote: | | cokehabit wrote: | | I think he was laughing more about the comment that reiser4 will show weaknesses in drives any more than any other filesystem. Like an "oogidy boogidy, reiser4 is so powerful it will break your HDD" which is laughable |
Mostly I was laughing at the amazing levels of outright wrongness that various Hans Ricer fanboys come up with... It's not your hardware, it's the crappy fs. |
And since you've used the "crappy fs" so much you know exactly how it works right?
Probably not.
 |
|
| Back to top |
|
|
Tiger683
Veteran
Joined: 08 Jan 2005
Posts: 1347
Location: Heffner's House
|
| Posted: Wed Apr 06, 2005 4:18 pm Post subject: |
|
|
| firephoto wrote: |
Probably not.
|
Ehm..... I wouldn't be so sure about that in HIS case.... Even though im not exactly same-minded.....
T
_________________
Retired gentoo user |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Wed Apr 06, 2005 4:23 pm Post subject: |
|
|
| firephoto wrote: | And since you've used the "crappy fs" so much you know exactly how it works right?
Probably not.
|
I've read the code. I've read the design docs. I found a dozen bugs in it before I even tried to compile it. So, uh, yeah, I know how it's supposed to work, and I know how it actually doesn't work. |
|
| Back to top |
|
|
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
| Posted: Wed Apr 06, 2005 6:14 pm Post subject: |
|
|
| ciaranm wrote: | | firephoto wrote: | And since you've used the "crappy fs" so much you know exactly how it works right?
Probably not.
|
I've read the code. I've read the design docs. I found a dozen bugs in it before I even tried to compile it. So, uh, yeah, I know how it's supposed to work, and I know how it actually doesn't work. |
And, I guess there was nothing that intrigued you? nothing you thought was cool?
I don't think you are wrong, as you obviously know more than I do - but as you know the way it is supposed to work - isn't that at least something you like? or is there really nothing new? |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Wed Apr 06, 2005 6:19 pm Post subject: |
|
|
| superstoned wrote: | | I don't think you are wrong, as you obviously know more than I do - but as you know the way it is supposed to work - isn't that at least something you like? or is there really nothing new? |
There's nothing innovative in reiser4 except for the way they make their metadata visible. Problem is, the way they make their metadata visible is the utterly broken part. That, combined with some seriously crap coding, makes the whole thing worthless. The whole "arbitrary metadata" thing was done to death thirty years ago on system/3[679]0, and it turned out to be not at all useful. |
|
| Back to top |
|
|
firephoto
Veteran
Joined: 29 Oct 2003
Posts: 1612
Location: +48° 5' 23.40", -119° 48' 30.00"
|
| Posted: Wed Apr 06, 2005 6:32 pm Post subject: |
|
|
| ciaranm wrote: | | superstoned wrote: | | I don't think you are wrong, as you obviously know more than I do - but as you know the way it is supposed to work - isn't that at least something you like? or is there really nothing new? |
There's nothing innovative in reiser4 except for the way they make their metadata visible. Problem is, the way they make their metadata visible is the utterly broken part. That, combined with some seriously crap coding, makes the whole thing worthless. The whole "arbitrary metadata" thing was done to death thirty years ago on system/3[679]0, and it turned out to be not at all useful. |
So if the code is so crappy then why is this crappy code making my "emerge syncs" much faster? If ext3 code is so superior (in case it is, I don't know that it is or isn't) then why have I lost data on every system that had a ext3 / ? I don't mean just some, I mean every box I ever installed some flavor of linux on that had ext3 for the main fs in use it always resulted in lost data at some point. Not just once, not a couple of times, but every time? Now this wasn't from graceful shut downs but under the same conditions with reiserfs zero problems occured. I guess what I'm saying is that I'll take the work of Hans' crappy coders if it means my data is safe through the hazards of every day desktop pc use over something that seems to only have the "tried and true" badge even though that doesn't match my real world situations. |
|
| Back to top |
|
|
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
| Posted: Wed Apr 06, 2005 6:55 pm Post subject: |
|
|
| ciaranm wrote: | | superstoned wrote: | | I don't think you are wrong, as you obviously know more than I do - but as you know the way it is supposed to work - isn't that at least something you like? or is there really nothing new? |
There's nothing innovative in reiser4 except for the way they make their metadata visible. Problem is, the way they make their metadata visible is the utterly broken part. That, combined with some seriously crap coding, makes the whole thing worthless. The whole "arbitrary metadata" thing was done to death thirty years ago on system/3[679]0, and it turned out to be not at all useful. |
you say it isn't faster for both large and small files? for multiple simultanous writes and reads? And it the fact it has atomic transactions, nothing new, every filesystem has it? and the fact that looking for files with a certain meta-data attribute is just as fast as looking for files in /etc or /home/user, just nothing new? plugins (if I understand it, its relatively easy to add compression, encryption etc)?
I'm not sure about these, but it seems to me reiser4 has these, and no other filesystem does. if I'm wrong, please point me to some docs about these things in other filesystems... maybe those using reiser4 are just brainwashed, and reiser4 had nothing new?
Look, you might be right, adding arbitrary meta-data may be the only thing. but why do they say one of reiser4's features? if all other features are just not new, like this one? or should I just believe you, rather than them? |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Thu Apr 07, 2005 12:16 pm Post subject: |
|
|
| superstoned wrote: | | you say it isn't faster for both large and small files? |
Thanks to the CPU hit, no, it's often slower in practical situations than some other filesystems.
| Quote: | | And it the fact it has atomic transactions, nothing new, every filesystem has it? |
Not every, of course. Just a whole load of others. Heck, atomic transactions date back at least to system/370.
| Quote: | | and the fact that looking for files with a certain meta-data attribute is just as fast as looking for files in /etc or /home/user, just nothing new? |
Heh. That's done in hardware on system/3[679]0, and done a hell of a lot better (see CKD). And looking for files that're, say, +x is just as fast as looking for files even on minixfs.
| Quote: | | plugins (if I understand it, its relatively easy to add compression, encryption etc)? |
Been doable via userspace for years. See all those loop-based projects out there. Sticking it in kernelspace is a misfeature. |
|
| Back to top |
|
|
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
| Posted: Thu Apr 07, 2005 12:51 pm Post subject: |
|
|
| ciaranm wrote: | | superstoned wrote: | | you say it isn't faster for both large and small files? |
Thanks to the CPU hit, no, it's often slower in practical situations than some other filesystems. |
Well, maybe on a i386, but for a 2000mhz+ processor (as most people use already, and more will in the future) the raw datatransfer is much more restricting than the processor usage... as long as it doesn't hamper interactivity, as it does - currently. but that's being worked on. and as Con already noted, the cpu-usage per transferred mb might not be much (if at all) higher than ext3, as it simply works faster... eg 10 mb/s, 1% cpu == 20 mb/s, 2% cpu...
btw I have to see the first proof its slower for anything, exept the interactive part (eg it sometimes seems to slow down the launching of applications, which seriously sucks, but again - its being worked on, I don't think its a fundamental problem, just implementation)
| ciaranm wrote: | | Quote: | | And it the fact it has atomic transactions, nothing new, every filesystem has it? |
Not every, of course. Just a whole load of others. Heck, atomic transactions date back at least to system/370. |
Yes, but they are very slow, performance wise, isn't it? or can ext3 (xfs, jfs) do them, without slowing things down? reiser4 at least can do it fast... and of course, ext3 simply does not have it. it might, in the future, but then we can only say they saw reiser4 had it, and copied it...
| ciaranm wrote: | | Quote: | | and the fact that looking for files with a certain meta-data attribute is just as fast as looking for files in /etc or /home/user, just nothing new? |
Heh. That's done in hardware on system/3[679]0, and done a hell of a lot better (see CKD). And looking for files that're, say, +x is just as fast as looking for files even on minixfs. |
my x86 can't do it in hardware, your's can? but ok, it might not be innovative, altough you say showing all files on a drive with +x on ext3 might be as fast as showing the same amount of files in one folder, I'm absolutely sure it is impractical - only reiserfs3 was able to handle such a thing (millions of files in 1 folder), and now - of course, reiser4.
Anyway, I wasn't only talking about this attribute, but more like what BeFS could do - show files with any kind of meta-data. other filesystems can do that (duh BeFS) but can they do it as fast as just ls /usr/bin? or is it slower? reiser4 *promises* its not any slower. I think that's interesting, as we (check the interview with Aaron Seigo about KDE 4.0) seem to be heading in the direction that the hierarchical filesystem is just one way to see data. it'd be bad if viewing it in another way (eg on mp3 tag) would be slower, OR require a database (which imho sucks, I'd love to see this in a filesystem without additional overhead of memory usage/diskspace/cpu usage a database has).
| ciaranm wrote: | | Quote: | | plugins (if I understand it, its relatively easy to add compression, encryption etc)? |
Been doable via userspace for years. See all those loop-based projects out there. Sticking it in kernelspace is a misfeature. |
Hmm, seen these. but can it be done fully transparant to all current applications, and again - without performance loss? Or don't you care about performance? you do, as you say it sucks reiser4 uses that much cpu...
btw: I hope you don't get mad, as I enjoy this argument, not only that, its very informative... its in no way meant to attack you or be rude! |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Thu Apr 07, 2005 1:08 pm Post subject: |
|
|
| superstoned wrote: | | Well, maybe on a i386, but for a 2000mhz+ processor (as most people use already, and more will in the future) the raw datatransfer is much more restricting than the processor usage... as long as it doesn't hamper interactivity, as it does - currently. but that's being worked on. and as Con already noted, the cpu-usage per transferred mb might not be much (if at all) higher than ext3, as it simply works faster... eg 10 mb/s, 1% cpu == 20 mb/s, 2% cpu... |
"reiser4 isn't a CPU hog except where it is. But we're working on that! Honest!"
| Quote: | | btw I have to see the first proof its slower for anything, exept the interactive part (eg it sometimes seems to slow down the launching of applications, which seriously sucks, but again - its being worked on, I don't think its a fundamental problem, just implementation) |
It's a design flaw. They forgot to consider the impact of figuring out where they'd hidden all the data.
Not their largest design flaw -- that would be not considering "doesn't lose data" to be worthy of inclusion on the planned features list.
| ciaranm wrote: | | Quote: | | And it the fact it has atomic transactions, nothing new, every filesystem has it? |
Not every, of course. Just a whole load of others. Heck, atomic transactions date back at least to system/370. |
Yes, but they are very slow, performance wise, isn't it? or can ext3 (xfs, jfs) do them, without slowing things down? reiser4 at least can do it fast... and of course, ext3 simply does not have it. it might, in the future, but then we can only say they saw reiser4 had it, and copied it...[/quote]
Slow? Uh, no.
| ciaranm wrote: | | Quote: | | and the fact that looking for files with a certain meta-data attribute is just as fast as looking for files in /etc or /home/user, just nothing new? |
Heh. That's done in hardware on system/3[679]0, and done a hell of a lot better (see CKD). And looking for files that're, say, +x is just as fast as looking for files even on minixfs. |
my x86 can't do it in hardware, your's can? but ok, it might not be innovative, altough you say showing all files on a drive with +x on ext3 might be as fast as showing the same amount of files in one folder, I'm absolutely sure it is impractical - only reiserfs3 was able to handle such a thing (millions of files in 1 folder), and now - of course, reiser4.[/quote]
Look up what system/3[679]0 are, what CKD is and then come back and rewrite the above so that it makes sense. Once you understand how CKD works, *then* you can start trying to make claims about reiser4 doing anything original.
| ciaranm wrote: | | Quote: | | plugins (if I understand it, its relatively easy to add compression, encryption etc)? |
Been doable via userspace for years. See all those loop-based projects out there. Sticking it in kernelspace is a misfeature. |
Hmm, seen these. but can it be done fully transparant to all current applications, and again - without performance loss? Or don't you care about performance? you do, as you say it sucks reiser4 uses that much cpu...[/quote]
Yup, loop mounts show up as a regular filesystem, and the performance isn't a problem either -- heck, sticking it in userspace means your entire kernel doesn't have to lock solid whilst it does nasty crypto stuff. The "no-one has SMP any more" 'argument' doesn't even apply here now that HT/SMT are so common. |
|
| Back to top |
|
|
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
| Posted: Thu Apr 07, 2005 2:12 pm Post subject: |
|
|
| ciaranm wrote: | | superstoned wrote: | | Well, maybe on a i386, but for a 2000mhz+ processor (as most people use already, and more will in the future) the raw datatransfer is much more restricting than the processor usage... as long as it doesn't hamper interactivity, as it does - currently. but that's being worked on. and as Con already noted, the cpu-usage per transferred mb might not be much (if at all) higher than ext3, as it simply works faster... eg 10 mb/s, 1% cpu == 20 mb/s, 2% cpu... |
"reiser4 isn't a CPU hog except where it is. But we're working on that! Honest!" |
It has already improved, and its mostly due to long codepaths, so I believe them.
| Quote: | | Quote: | | btw I have to see the first proof its slower for anything, exept the interactive part (eg it sometimes seems to slow down the launching of applications, which seriously sucks, but again - its being worked on, I don't think its a fundamental problem, just implementation) |
It's a design flaw. They forgot to consider the impact of figuring out where they'd hidden all the data. |
well, if that's the case, it won't be fast. so as proof will come saying its slow, no-one will bother using it, right? future will tell who's right on this (as I think it IS faster than other filesystems, and most benchmarks I saw say it is, as do most people that tried it).
| Quote: | Not their largest design flaw -- that would be not considering "doesn't lose data" to be worthy of inclusion on the planned features list. | Quote: | yeah, why isn't that on ext3's featurelist? I lost data with it three times on a row, so it clearly sucks. My data isn't safe on ext3, actually I expect reiser4 (with its totally un-innovative atomic transactions) to become safer as ext3 is.
| Quote: | | ciaranm wrote: | | Quote: | | And it the fact it has atomic transactions, nothing new, every filesystem has it? |
Not every, of course. Just a whole load of others. Heck, atomic transactions date back at least to system/370. |
Yes, but they are very slow, performance wise, isn't it? or can ext3 (xfs, jfs) do them, without slowing things down? reiser4 at least can do it fast... and of course, ext3 simply does not have it. it might, in the future, but then we can only say they saw reiser4 had it, and copied it... |
Slow? Uh, no. |
atomic transactions are just as fast as non-atomic? hmmm. I wonder why ext3, xfs and jfs don't have them.
| Quote: | | ciaranm wrote: | | Quote: | | and the fact that looking for files with a certain meta-data attribute is just as fast as looking for files in /etc or /home/user, just nothing new? |
Heh. That's done in hardware on system/3[679]0, and done a hell of a lot better (see CKD). And looking for files that're, say, +x is just as fast as looking for files even on minixfs. |
my x86 can't do it in hardware, your's can? but ok, it might not be innovative, altough you say showing all files on a drive with +x on ext3 might be as fast as showing the same amount of files in one folder, I'm absolutely sure it is impractical - only reiserfs3 was able to handle such a thing (millions of files in 1 folder), and now - of course, reiser4. |
Look up what system/3[679]0 are, what CKD is and then come back and rewrite the above so that it makes sense. Once you understand how CKD works, *then* you can start trying to make claims about reiser4 doing anything original. |
well, I can't find much, with google at least. I don't have much knowledge, so it has to be an easy read... anyway, CKD can't be choosen as filesystem under linux, I guess? or can it? if so, why don't we all use it, if its so much better...
hey, it doesn't matter if other platforms have things reiser4 has, as long as reiser4 is the only thing that can bring these to MY computer (x86)... no-one uses a non x86 anyway
(yeah, I know, many do, but still just a few %s, 95% is x86 or x86-64. add a few ppc, and there is 1 or 2 % left! who cares if they have filesystems x86 does not? if these filesystems where cool and free software, they'd be ported by now. looks like they sucked, OR where not free...) |
|
| Back to top |
|
|
ciaranm
Retired Dev
Joined: 19 Jul 2003
Posts: 1719
Location: In Hiding
|
| Posted: Thu Apr 07, 2005 2:35 pm Post subject: |
|
|
...so what it basically boils down to is that reiser4 is not innovative, but you're calling it 'innovation' because it's the first time someone's done it on Linux.
So if Microsoft were to add proper CSS2 support to IE, would that be innovating? |
|
| Back to top |
|
|
feld
Guru
Joined: 29 Aug 2004
Posts: 593
Location: WI, USA
|
| Posted: Thu Apr 07, 2005 4:19 pm Post subject: |
|
|
I dont know what to say. I liked what I experienced with reiser4. I ran it from XMAS until yesterday. It was the fastest FS I've ever used. Does it have bugs? I dont know. I never experienced any. No lockups either. It may have incompatibilities with software, hardware, and other archs. Does this mean its junk? No! not at all! It needs guidance. It needs help.
Its reputation is going to be from user experience and developers insights. This filesystem will lose all chance of hope if all everyone does is slander it. Admit it has faults, but at least admit that under the correct guidance it COULD be something great. Maybe then more community members will take note and help work on it. Maybe then all major bugs will disappear, and it will be slowly available on other archs and every hardware setup. Remember, this is GPL... so you guys CAN do something about it.... we just need more people to care.
-Feld
_________________
< bmg505> I think the first line in reiserfsck is
if (random(65535)< 65500) { hose(partition); for (i=0;i<100000000;i++) print_crap(); } |
|
| Back to top |
|
|
firephoto
Veteran
Joined: 29 Oct 2003
Posts: 1612
Location: +48° 5' 23.40", -119° 48' 30.00"
|
| Posted: Thu Apr 07, 2005 4:59 pm Post subject: |
|
|
| feld wrote: | I dont know what to say. I liked what I experienced with reiser4. I ran it from XMAS until yesterday. It was the fastest FS I've ever used. Does it have bugs? I dont know. I never experienced any. No lockups either. It may have incompatibilities with software, hardware, and other archs. Does this mean its junk? No! not at all! It needs guidance. It needs help.
Its reputation is going to be from user experience and developers insights. This filesystem will lose all chance of hope if all everyone does is slander it. Admit it has faults, but at least admit that under the correct guidance it COULD be something great. Maybe then more community members will take note and help work on it. Maybe then all major bugs will disappear, and it will be slowly available on other archs and every hardware setup. Remember, this is GPL... so you guys CAN do something about it.... we just need more people to care.
|
There's probably some history hidden away somewhere that would reveal what Hans Reiser has done in the past to piss off a bunch of the kernel devs but what it is exactly I don't know but ever since he started with reiser3 and working with the kernel for it he's faced a battle. I've read some old usenet posts and mailing lists stuff from the original reiserfs era and the arguments against it are strikingly similar to the modern arguments against reiser4 and I even spotted the same names a few times if I remember correctly. I guess he whizzed in someones cheerios and they're still ticked off. |
|
| Back to top |
|
|
rhill
Retired Dev
Joined: 22 Oct 2004
Posts: 1629
Location: sk.ca
|
| Posted: Fri Apr 08, 2005 5:37 am Post subject: |
|
|
i honestly don't know the history, but my personal opinion of Namesys as an entity was formed by this page:
http://www.namesys.com/support.html
that just doesn't sit right with me. every time i see them argue something like "Don't move reiser4 into vfs, use reiser4 AS the vfs."[1] i have to wonder what their motivation is - free open source or cash in hand. your personal views may/probably differ and that's cool, i'm just offering a look at mine. i don't think their motives should keep them from becoming a part of the kernel, once the issues with R4 are resolved, but i think they need to be kept in mind.
[1] http://lwn.net/Articles/100154/
_________________
by design, by neglect
for a fact or just for effect |
|
| Back to top |
|
|
Suicidal
l33t
Joined: 30 Jul 2003
Posts: 959
Location: /dev/null
|
| Posted: Fri Apr 08, 2005 6:54 am Post subject: |
|
|
| feld wrote: | | but at least admit that under the correct guidance it COULD be something great. |
I totally agree but having reiser4 flake out on me is not good for business . There is nothing worse than configuring ACL's on routers and pull out your lappy to test them with tcptraceroute to find the FS has gone south (luckily it was still in the shop on the bench). I also agree people should test the FS but not everyone.
I currently use reiserfs 3.6 on all of my systems and it has been the most stable and almost the most responsive FS I have used on any OS bar none. But still little flake outs like this is proof enough that its not production stable and people that value thier data should steer clear until it matures. |
|
| Back to top |
|
|
superstoned
Guru
Joined: 17 Dec 2004
Posts: 432
|
| Posted: Fri Apr 08, 2005 8:33 am Post subject: |
|
|
| Suicidal wrote: | | feld wrote: | | but at least admit that under the correct guidance it COULD be something great. |
I totally agree but having reiser4 flake out on me is not good for business . There is nothing worse than configuring ACL's on routers and pull out your lappy to test them with tcptraceroute to find the FS has gone south (luckily it was still in the shop on the bench). I also agree people should test the FS but not everyone.
I currently use reiserfs 3.6 on all of my systems and it has been the most stable and almost the most responsive FS I have used on any OS bar none. But still little flake outs like this is proof enough that its not production stable and people that value thier data should steer clear until it matures. |
if the possibility of dataloss would make a filesystem not production stable, no filesystem is stable. I've had dataloss with all filesystems I used, so I'd say trust none, and make backups... |
|
| Back to top |
|
|
cokey
Advocate
Joined: 23 Apr 2004
Posts: 3355
|
| Posted: Fri Apr 08, 2005 10:51 am Post subject: |
|
|
| Suicidal wrote: | | feld wrote: | | but at least admit that under the correct guidance it COULD be something great. |
I totally agree but having reiser4 flake out on me is not good for business . There is nothing worse than configuring ACL's on routers and pull out your lappy to test them with tcptraceroute to find the FS has gone south (luckily it was still in the shop on the bench). I also agree people should test the FS but not everyone.
I currently use reiserfs 3.6 on all of my systems and it has been the most stable and almost the most responsive FS I have used on any OS bar none. But still little flake outs like this is proof enough that its not production stable and people that value thier data should steer clear until it matures. |
couldn't have put it better myself.
Although it shows promise, it shouldn't be let anywhere near anything that is not a test machine and therefore SHOULDN'T be classed alongside such filesystems as jfs, ext3, xfs and (ciaranm's going to hate me) reiser 3.6
_________________
https://otw20.com/ OTW20 The new place for off the wall chat |
|
| Back to top |
|
|
Archangel1
Veteran
Joined: 21 Apr 2004
Posts: 1212
Location: Work
|
| Posted: Fri Apr 08, 2005 11:37 am Post subject: |
|
|
That's not _that_ bad - the software is free, but if you want an answer from them you have to pay for it.
Well $25 isn't bad for a serious problem - especially given that they would then help you, rather than just laughing and saying "RTFM" or "that's what you get for using unstable software".
It certainly seems reasonable after having talked to Microsoft tech support - I bought a copy of Win2k Pro a few years back. At some point IIS stopped working. I called them up, and after explaining to the guy what IIS was he told me I'd be charged AU$300/hr and connected with a tech in Sydney.
I'll take Mr Reiser over that any day.
Sure, it's not as nice as totally free support, but who's offering that? The support on these forums is great, but I have posted things that haven't been responded to because, presumably, nobody knew the answer or could be bothered replying. At least when you pay your $25 you do get a response.
And it's not like you have to pay - or even use it. Nobody's forcing you to.
| dirtyepic wrote: | that just doesn't sit right with me. every time i see them argue something like "Don't move reiser4 into vfs, use reiser4 AS the vfs."[1] i have to wonder what their motivation is - free open source or cash in hand. your personal views may/probably differ and that's cool, i'm just offering a look at mine. i don't think their motives should keep them from becoming a part of the kernel, once the issues with R4 are resolved, but i think they need to be kept in mind.
[1] http://lwn.net/Articles/100154/ |
There are obviously a lot of good points both ways.
It really strikes me that he's made a filesystem which simply does a lot of things other filesystems don't, and is being told it won't go into the mainline kernel because it's different. By the same argument, you could exclude all sorts of features from the kernel.
And I think he's got a point that it's more advanced than Apple's or Microsoft's efforts, and we should be building on that so Linux remains the most advanced OS around. It's nice to think at the moment that it's technically competitive with the alternatives; I don't think playing catchup once Tiger and Longhorn come out would be a lot of fun.
I agree that that vfs quote is a bit disturbing though - Linux is bigger than just one filesystem.
I am a bit biased on this topic - having not lost any data to reiser4 in the last nine months or so I don't see what the big problem people have with it is. Of course I haven't really thrashed it either, I guess it could be getting a lot more work than it is.
_________________
What are you, stupid? |
|
| Back to top |
|
|
brantgurga
Tux's lil' helper
Joined: 30 Dec 2003
Posts: 75
Location: Indianapolis, IN
|
| Posted: Fri Apr 08, 2005 4:11 pm Post subject: |
|
|
| Regarding the $25, it really applies to support issues. I reported a possible problem caused by Windows overwriting the partion tables at a clean reinstallation and then using cfdisk to restore the original partition layout. reiser4 was the only partition not recovered. I didn't need support on the issue since I didn't need any data recovered; I simply reported it and still got a response regarding some things to check next time it happens like fsck before windows repartition, same starting and ending sectors, and some other items. I think they were pretty reasonable. Hans really seems to want to have a reputation for good service, even if it isn't paid for. If I actually needed to recover data, then I would have paid the $25. |
|
| 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
|
|
Unsupported Software
