confusion with reinstall of seawolf
Fuzzy
I just reinstalled, (using the "server" option of text install).
It appears whatever the supplied kernal is, it doesn't support
ext3. the -j option is missing from the fsprogs also.
what do I need to do to get support for ext3?
does the current version of roswell's ISOs
work correctly, the version I have fails
on a perl package? I tried repeatedly
to download new versions and recreate the CDs.
fuz
_______________________________________________
Redhat-install-list mailing list
Redhat-in...@redhat.com
https://listman.redhat.com/mailman/listinfo/redhat-install-list
Kalum / Grendel
>
> I just reinstalled, (using the "server" option of text install).
> It appears whatever the supplied kernal is, it doesn't support
> ext3. the -j option is missing from the fsprogs also.
>
> what do I need to do to get support for ext3?
To support that shit ext3, I call it shit because there are better file
systems, for example XFS, to give you a idea, my boot time (after init, to
come to the prompt) dropped down by 25% after I installed XFS,its such a
good file system, utterly fast in fact it is the fastest, it is even
faster than non journalling file systems like ext2.
Anyway about installing that piece of beta software called ext3,
You will need a kernel
1. Get the alan cox (ac) series, linus's vanilla kernels dont support
ext3, quite rightly infact as ext3 is still beta, has enormous problems
with quota support, crashes under high loads :(
Because alan cox works for redhat and redhat decides that its time that we
use such shit, you will be pleased to find out that the ac-series support
it by default, so you should be able to run a make x config and just add
support for it the usual way.
3. updated version of the tune2fs utitlty, they as part of the e2fs-progs
package.
All you have to do is just to add a journal to the ext2 file system,
tune2fs -j /dev/hdXX
replace the XX with the values appropriate for your patition, now you are
ready to go, make sure you fstab refers to the partition type as auto, and
thats all, reboot and enjoy the slow down in speed.
<http://www.uow.edu.au/~andrewm/linux/ext3/ext3-usage.html>
Personnaly I think ext3 is so immature and unstable (its really a
ext2 with a journal) that it actually is risky running it.
In terms of speed it is nowhere near XFS (which is run on SGIs high
performance machines), and it also does poorly in comparison with
reiserfs.
In short get a ext3 supported kernel, and the latest e2fs progs and just
add the journal, set the file system type to auto and notice what a
sucker ext3 is.
What redhat should be doing is to see us running XFS (linux on steroids),
or reiserfs (time tested) rather than forcing such beta shit which even
linus does not include in his kernel.
Best Wishes,
Grendel
--
.---------------------.---------------------.----{)--.
| /"__ ._ _ _ _| _ |`-. ka...@delrom.ro .-'(]__/|| |
| \__/ | (-'| |(_|(-'l_ `-===============-' [_] .-: |
`--------------------------------------------/|\/| |-'
all your chix are belong to us.
Kalum / Grendel
This is a short discussion about my experience with various journalling
file systems with linux and what I think of them, I have tried to be
balanced as possible and speak from a practical viewpoint, as I think it
should be, we are not interested in theory we are only interested in how
it works in practice.
I will be considering ext2 (although it is non journalling I take it here
as it is the de facto file sstem for linux), ext3, XFS, and resierfs.
Ext2
=========================================
What is it?
It is a non journalling file system, the defacto standard file system for
linux for a long time.
Benefits:
Extensive testing, reliable except after crashes data loss is quite
common, it is very fast, but nowhere as fast as a non journalling fs
should be, as despite being slowed down by journalling XFS beats it
in speed by a long way.
Currently ext2 is one of the fastest, beaten only by XFS. Reiserfs is
slower than ext2 for general usage.
Problems:
After a power off, or crash, you have to fsck the system, and a file
system check is very slow, for example fscking a 3gb partition takes about
4 minutes, so people with 40gb partions should not run it, as it will be
so slow to bring up after a crash.
ext3
====================================================================
A journalling file system which is a improvement over ext2, this is going
to be used by redhat as there default fs in the upcoming rh 7.2 release.
Benefits:
Since it is a journalling file system no need to wait for long fscks after
crashes or power outages.
That is the only benefit it has as its disadvantages far outweigh the
above benefit.
Problems:
These are many since ext3 is still beta software which has not reached
even the 1.0 stage. Most notable are that it does not support quotas
properly and also that it crashes under high loads, which limit its
viaility in a production quality environment. It is much much slower than
ext2 and thus offers no signifcanct speed improvement over ext2. It is
also incompatible with linus's kernels.
In case you dont beleive me, here is a list of the changes made to a ext3
release this very morning,
ext3 0.9.12
* A fix for an oops in directory readahead on empty (i.e., deleted)
directories.
Grendel: Wow, so this is the so called stable file system that redhat
plans to include in their new release for production quality code
* handling the new S_NOQUOTA flag for 2.4.10+ quota code
Grendel: Still the quota support is very flaky, and this shows it.
* reverting i_truncate_sem vs. i_sem lock ordering to original design to
fix an obscure deadlock (but will deadlock 2.4.11's shmem.c; it's safe
with 2.4.10-ac11's shmem.c).
Grendel: ahh.. deadlocking code (se my above statements about ext3 locking
up under high loads), hmm of course it "might" work with ac kernel series,
infact ext3 has serious problems working with any linux kernel.
So to put it bluntly ext3 is a piece of shit, which should not be used in
a production quality environment yet, and this confirms my (and also Hans
Reisers) belief that redhat is a company which just bundles software
without thinking much about it.
I appeal to you not to use ext3 as it offers no real benefits over ext2,
apart from the fast uptime after a crash.
Also ext3 runs that fscking kjournald, a journalling kernel thread,
something which no other journallign file systems sems to need, this
kjournald consumes .5% of my cpu time on my 500 mhz athlon, which is way
to excessive.
XFS
=====================================
A relatively new entry to the linux market, but which has been in SGI's
irix for over 10 years, and thus is the most tested fs out there.
Positives:
Awesome speeds, it is the fastest fs out there whcih has been confirmed by
numerous benchmarks, and my own experiences confirm this, for example
my boot time after the init starts to the apperance of the prompt after
loading a lot of apps that need disk access like apache, and squid was
reduced by 25% over ext2's booting.
This produces a stunning 20 Mb/sec *sustained* data transfer rates over my
old 5200 rpms quantum firebal ide drive.
Extremely reliable, I run my root fs and every fs on this, I have had
several power cuts, the recovery performed in 1 second was perfect, I
have never had any data loss. Stability too is extremely impressive.
Disadvantages:
None, except the fact that you have to patch your kernel with SGI's
patches to get it working. But it is worth it. Also there are iso's
available which add xfs support to the base redhat iso's.
Reiserfs
===================================================================
A journalling file system popularised by SUSE.
Advatages:
Journalling thus fast recovery, used by suse as a production quality
environment with no problem whatsoever.
Included as part of the kernel patches.
Reliable, I have used it and it is very stable but nowhere as fast as XFS.
Disadvantages
=================================================
Slow: Slower than ext2, however faster than that so called "file system"
ext3.
Immature, its one of the youngest file systems out there. XFS and JFS have
had more experience than this.
That brings to a end my thoughts on journalling file systems, my heartiest
recommendation would be for you to try XFS and see the change, when I
first installed it on our own list member Steve Pardee's insitance I was
skeptical, till I actually saw it in action, and I thought "man this cant
be hapenning, its so fast". It took me only 2 days after the intro to
covert all my file systems to XFS such was by awe, and I have remained on
XFS ever since.
To install XFS, please visit
<http://rpmfind.net/linux/SGILinux/patches/>
And download the corresponding patch for your linus's kernel, please
ignore the patches with the "-cvs" tag in them.
Then patch the kernel and in the file systems section enable support for
XFS.
Now that you have a running kernel which supports XFS, you need to
xfsprogs package to create a xfs file system.
<ftp://oss.sgi.com/projects/xfs/download/cmd_tars/>
<ftp://oss.sgi.com/projects/xfs/download/cmd_tars/xfsprogs-1.3.7.src.tar.gz>
Is more than enough for the moment, just compile and install it.
Then use
/sbin/mkfs -t xfs /dev/hdXX
and then mount it as usual with the type set to xfs.
mount -t xfs /dev/hdXX /mnt/xfs
and have fun.......
Fuzzy
Kalum / Grendel wrote:
>
> XFS
> =====================================
Mark W. Knecht
Thank you for continuing to answer every email with profanity.
With little regard,
Mark
Ryan Waldron
> Grendel,
> Thank you for continuing to answer every email with profanity.
On the plus side, she gives people a good reason to eventually go and
learn how to filter and delete incoming messages by sender with tools
like procmail (http://www.procmail.org), which is one of the Truly
Great Tools available to Linux (/BSD/Un*x) users. Necessity is the
mother of invention, and the father of learning.
:)
--
Ryan Waldron ||| http://www.erebor.com ||| r...@erebor.com
"The web goes ever, ever on, down from the site where it began..."
Steve Pardee
Hat ISOs. Very well done. It has been a while since I have done one of
these installs, but if I rember correctly, the only gotcha I recall is
that if you let the install auto-create your partitions, it will by
default create ext2, so you will have to semi-manually set your
partitions up and then override the default ext2 setting and select XFS
as your file system. I would be very surprised if you do not see a
noticable increase in disk performance. Have fun.
Cheers,
Steve Pardee
Fuzzy wrote:
>
> where can one find the ISO files ?
>
> Kalum / Grendel wrote:
> >
>
> > XFS
> > =====================================
> >
>
> >
"Reality is merely an illusion, albeit a persistent one."
--Albert Einstein
Steve Pardee
ftp://ftp.thebarn.com/SGI/RH7.1-SGI-XFS-1.0.1/images/
--Steve Pardee
Fuzzy wrote:
>
> where can one find the ISO files ?
>
> Kalum / Grendel wrote:
> >
>
> > XFS
> > =====================================
> >
>
> >
"Reality is merely an illusion, albeit a persistent one."
--Albert Einstein
_______________________________________________
Stephen Walton
--
Stephen Walton, Professor of Physics and Astronomy,
California State University, Northridge
stephen...@csun.edu
On Mon, 15 Oct 2001, Steve Pardee wrote:
> Oops, you can download the ISO from
> ftp://ftp.thebarn.com/SGI/RH7.1-SGI-XFS-1.0.1/images/
>
> --Steve Pardee
>
_______________________________________________
Steve Pardee
where I first learned about Journaling File Systems. So last winter and
early spring I started experimenting with XFS and another guy on our
team experimented with Reiserfs. That is my only 2 reference points
other than anecdotal stuff people have told me. XFS is faster than
Reiserfs and others have told me there have been performance tests done
and that XFS is the fastest; Grendel who helps out people on this
newsgroup might be able to point you to a report. We have been running
XFS pre 1.0 on some of our production servers since March and they have
proven to be equal to or more reliable than ext2 based servers. The SGI
ISO in very cleanly integrated into the Red Hat install. It requires
the two Red Hat 7.1 ISOs and is the install is no more difficult than
installing Red Hat. If you have further questions, please let me know.
I'll be happy to share what I have with you.
Sincerely,
Steve Pardee
"Reality is merely an illusion, albeit a persistent one."
--Albert Einstein
_______________________________________________
Kalum / Grendel
> How does IBM JFS compare to XFS in performance and ease of install?
Currently JFS's problems are that
1. Its slow, even the developers in IBM's JFS site claim that it is slower
than ext2, with the comment "A journalling file system will never be as
fast as a non journalling one" in reply to the question as to how fast
their file system is, so this disparaging comment by ther own
developers made me not try it.
2. Serious stability problems, I saw this thread on where the mandrake
developers were seriously considering not adding JFS to their mandrake
8.1, due to prolems with the stability testing, but it may be fixed now.
As I said I have not tried this out, and what I am saying is from comments
by developers themselves and people who have tried to use it.
Best Wishes,
Grendel
--
.---------------------.---------------------.----{)--.
| /"__ ._ _ _ _| _ |`-. ka...@delrom.ro .-'(]__/|| |
| \__/ | (-'| |(_|(-'l_ `-===============-' [_] .-: |
`--------------------------------------------/|\/| |-'
all your chix are belong to us.
Ryan Waldron
> How does IBM JFS compare to XFS in performance and ease of install?
>
Here are a few links that might be of interest to you. The first is
an interview with some of the developers behind ReiserFS, JFS, and
XFS, talking about their respecitve projects:
http://www.osnews.com/story.php?news_id=69
IBM has a very details "Red Paper" in which (partway down) they talk
about scalability and performance for JFS, XFS, and ReiserFS:
http://www.redbooks.ibm.com/redpapers/pdfs/redp0121.pdf
This message:
http://oss.sgi.com/projects/xfs/mail_archive/0107/msg00046.html
mentions the lack of an RH-based installer for JFS, which XFS does
appear to have (I haven't used either of them).
SGI's XFS project page:
http://oss.sgi.com/projects/xfs/index.html
is fairly good, as is IBM's JFS page:
http://oss.software.ibm.com/developerworks/opensource/jfs/
There is a large amount of material there that might be useful to you,
including a tutorial on JFS (which should at least walk you through
the process of installing it, which might give you some idea).
This is by no means exhaustive, but perhaps it's more useful than the
breathless "Ooh, JFS sucks 'cuz I think I read an article by some
developer and I think I remember maybe he said it was slow or
something and I don't know but it sucks and you shouldn't use it" sort
of answer you seem to have gotten from one of our regulars here. :)
--
Ryan Waldron ||| http://www.erebor.com ||| r...@erebor.com
"The web goes ever, ever on, down from the site where it began..."
_______________________________________________
Mike Karmindro
someone's tip I kinda fixed it. right now everytime I log in (via rlogin) i
have to do an 'export DISPLAY=....." and then it works. Also I can't figure
out how to increase the font size! In the emacs drop down menues there is
nothing to do this (amazingly).
Can someone help me fix these two things so it loads up every time so I
don't have to manually export and increase my font size? thx.
i thought i read somewhere about an .emacs file that has settings, but i
can't seem to find that either...
Steve Pardee
before. I still need to re-read some of it too. It was very
enlightening. I have been using XFS for almost a year and the reasons I
selected it and continue be enthusiastic are: 1. It is as easy to
install as Red Hat 2. It is faster by a noticable margin than the other
journaling file sustems and 3. It is rock solid in terms of
reliability. I am hoping someone will come along with a superior
solution and prove my decision wrong, but ALL the feed-back I have
received so far over the last 9 mos. or so have gone the other way. If
you have anymore feedback, please copy me on it if you would please.
Cheers,
Steve Pardee
Ryan Waldron wrote:
>
> On Mon, 15 Oct 2001, Stephen Walton wrote:
>
> > How does IBM JFS compare to XFS in performance and ease of install?
> >
>
> _______________________________________________
> Redhat-install-list mailing list
> Redhat-in...@redhat.com
> https://listman.redhat.com/mailman/listinfo/redhat-install-list
"Reality is merely an illusion, albeit a persistent one."
--Albert Einstein
_______________________________________________
Kalum / Grendel
> On Mon, 15 Oct 2001, Mark W. Knecht wrote:
>
> > Grendel,
> > Thank you for continuing to answer every email with profanity.
>
> On the plus side, she gives people a good reason to eventually go and
Sorry to disappoint you but I am a "he".
> learn how to filter and delete incoming messages by sender with tools
> like procmail (http://www.procmail.org), which is one of the Truly
> Great Tools available to Linux (/BSD/Un*x) users. Necessity is the
> mother of invention, and the father of learning.
My language may be slightly out of order, but my advice is well worth not
blocking my postings all together, dont you think so Ryan? :)
Besides dont you watch movies where words like fsck shit etc, are quite
commonly used, I am surprised that you dont block/censore them then...
Anyway back to helping new users...
Best Wishes,
Grendel
--
.---------------------.---------------------.----{)--.
| /"__ ._ _ _ _| _ |`-. ka...@delrom.ro .-'(]__/|| |
| \__/ | (-'| |(_|(-'l_ `-===============-' [_] .-: |
`--------------------------------------------/|\/| |-'
all your chix are belong to us.
Kalum / Grendel
> Grendel,
> Thank you for continuing to answer every email with profanity.
Lets put it this way, I must admit that my comments are somewhat
unrestricted, but I make them with my own personal experience and with the
desire to share it with others.
Lets get to the bottom of the matter, do you think ext3 is ready to be
used as production line file system, all I need is a yes or no, and a
justification for it if you say yes.
The only reason that Rh wants to get out this beta thrash ext3 is because
it is being developed by redhat, and as usual RH doesnt care what it
shoves on to cd's, just look at the gcc-2.96 fiasco, gcc-2.96 does not
compile MMX codecs, programs like mplayer, aviplay all fail misreably
thanks to RH's idiocy :(
I make no reservations about calling a spade a spade in the strongest
possible sense, my criticism of ext3 are very valid (just read the new
change log for it ) and calling it a piece of shit is putting it mildly
when you think that RH is going to bring it into the frontline such a beta
immature file system, when there are btter more tested file systems out
there.
Ralph Slooten
experiences, knowledge and so on. If you wish to delete all posting from
Kalum, then be my guest, but on the other hand, you may learn something :-)
And I would say that the "father of learning" would be "listen and learn",
and not ridicule, mock and make really pointless remarks.
Kalum has helped me out numerours times in the past, and will too in the
future. It so happens that he has a vast knowledge of Linux, and yes, he
can get "emmotionally" involved sometimes, buy hey, Linux is a religion.
But again, if you would like to configure your procmail to block his
reples, go right ahead, at least then we wouldn't get these stupid
replies each time :-)
Keep up the good work of helping people Kalum, even if it is just your
opinion, at least you take the time to share it with us.
Greetings
Ralph
On Tue, 16 Oct 2001, Kalum / Grendel wrote:
> On Mon, 15 Oct 2001, Ryan Waldron commented thusly,
> > learn how to filter and delete incoming messages by sender with tools
> > like procmail (http://www.procmail.org), which is one of the Truly
> > Great Tools available to Linux (/BSD/Un*x) users. Necessity is the
> > mother of invention, and the father of learning.
>
> My language may be slightly out of order, but my advice is well worth not
> blocking my postings all together, dont you think so Ryan? :)
>
> Besides dont you watch movies where words like fsck shit etc, are quite
> commonly used, I am surprised that you dont block/censore them then...
>
> Anyway back to helping new users...
>
> Best Wishes,
> Grendel
--
ICQ: 25543458 Homepage: http://www.axllent.linux-dude.net/
PGP Public Key: http://prozilla.delrom.ro/RalphSlooten.asc
Kalum / Grendel
> This is by no means exhaustive, but perhaps it's more useful than the
> breathless "Ooh, JFS sucks 'cuz I think I read an article by some
> developer and I think I remember maybe he said it was slow or
> something and I don't know but it sucks and you shouldn't use it" sort
> of answer you seem to have gotten from one of our regulars here. :)
Ryan, instead of babbling like a baboon in heat, please visit the JFS FAQ
itself,
<http://www.moelabs.com/modules.php?op=modload&name=FAQ&file=index&myfaq=yes&id_cat=1&categories=JFS
>,
and see the developers lame responses to the questions, they know that
JFS is slow and they are trying their best to cover it, for example...
· How fast is JFS compared to ext2?
Folks, there is no way around it: journaling is slower inherently compared
to non-journaling file systems. Some code has to make for the added
security and that code slows down I/O operations. The good news is that
not even ext2fs is as fast as it could be theoretically, so JFS tries to
improve things by being fast at basic I/O operations, be they for
journaling or for data.
Grendel: I am sure that such answers are very inspiring to a prospective
user of JFS, the bottom line is JFS is slow, even slower than ext2.
About stability,
<http://mt.mandrake.org/mcwn/2001/08/27/>
Please not that in the end the developer mentions..
"JFS-1.0.3 (well we still have many stability problems with it... our QA
team, after an intensive run of stress tests, adviced us to remove the
easy availability for JFS partitions during install; concerning journaling
filesystems, you still may choose from Reiserfs and Ext3)"
Ryan, I gain nothing from helping people, I do it as a sense of duty, I
try to do it even with the minimum knowledge that I have, but what ever I
say is allways substantiated, so rather than directing personal attacks at
me like the above which show that you are a imbecile please help people
instead.
So in a summary, JFS still has a lot of problems and does not give any
impetus currently as to use it over ext2.
Best Wishes,
Grendel
--
.---------------------.---------------------.----{)--.
| /"__ ._ _ _ _| _ |`-. ka...@delrom.ro .-'(]__/|| |
| \__/ | (-'| |(_|(-'l_ `-===============-' [_] .-: |
`--------------------------------------------/|\/| |-'
Prozilla: all your files are belong to us.
Ryan Waldron
> On Mon, 15 Oct 2001, Ryan Waldron commented thusly,
> > On the plus side, she gives people a good reason to eventually go and
>
> Sorry to disappoint you but I am a "he".
My sincere apologies on that one. I thought I remembered some time
ago (many months, maybe a year or more) I had read a message from you
correcting someone for referring to you as 'he', and had made a point
to try to remember it correctly. I meant no slight by misrepresenting
your gender.
> My language may be slightly out of order, but my advice is well
> worth not blocking my postings all together, dont you think so Ryan?
> :)
Probably I'm not the best one to ask, as you know I disagree quite
pointedly with both your belief and your mode of expression of it on
at least one prominent software package. :)
However, since you asked, however seriously, I'll explain what my
problem is with your approach. It's not that you're not helpful, or
that you're not generally forthcoming with useful information. It's
that you're flatly and unrelentingly dismissive of software that, for
whatever reason, *you* don't like.
Your language indicates not only a decision on your part that it
doesn't work for you, but that it furthermore shouldn't work for ANY
halfway sane person (probably why you called me an imbecile in a later
post tonight).
Since this is a somthing of a Linux newbie list (though not entirely),
your loud and repeated blanket condemnations of rather influential and
potentially important software may have the undesirable effect of
limiting people's exposure to software that they might learn something
from.
It's one thing to say, "I really, really hate this software for this
reason and that." It's another to say, repeatedly and at every
opportunity, that "this or that piece of software is complete
[excrement], and only used by imbeciles, and was written by brain-dead
monkeys, and you'd have to be both ugly AND stupid to ever even
consider it," which is pretty much the impression you give on a
variety of subjects.
You certainly have the right to say it. I likewise have the right to
disagree with or satirize such judgmental and derisive (and, in my
opinion, occasionally very poorly informed and weakly reasoned)
pronouncements. I do that not specifically to ruin your day, which I
don't think it does. I believe that you, like me, have been around
long enough that you don't shed tears when someone attacks something
you said. Rather, I do it to just let it be known that your *opinion*
is just that, and does not represent the "received wisdom of the
community" on that topic, at least not without dissent.
And that's part of what a discussion list is for, right? To see more
than one side, and to hear the voices of opposition at times?
--
Ryan Waldron ||| http://www.erebor.com ||| r...@erebor.com
"The web goes ever, ever on, down from the site where it began..."
_______________________________________________
Thomas Dodd
Kalum / Grendel wrote:
> · How fast is JFS compared to ext2?
>
> Folks, there is no way around it: journaling is slower inherently compared
> to non-journaling file systems. Some code has to make for the added
>
> Grendel: I am sure that such answers are very inspiring to a prospective
> user of JFS, the bottom line is JFS is slow, even slower than ext2.
Which is the truth. a journled FS will be slower than a non-journled
one if both are well designed and optimized. It has to be, There's
at least 2 extra I/O ops, write the journal that a transaction is
scheduled, and update the journal when the transaction is finished.
The code to monitor the journal is minor, compared to the I/O,
but could be a factor between 2 journaling FS's.
> So in a summary, JFS still has a lot of problems and does not give any
> impetus currently as to use it over ext2.
Other than the fact that it *IS* a journaling FS, which ext2 is not.
What journaling file system is faster than ext2 when it is actually
writing to a journal?
Is the journal on a seperate disk?
Most users today only have 1 disk.
Was the test IDE or SCSI?
If IDE, was it the integrated controler or PCI addon?
Both drives masters? Do either have slaves on the channel?
What was the data used in the test?
-lots of small files in 1 directory
-lots of directories with a few small files in each
-the above 2 with large files
-mixtures of the above
-Huge files sequential reads/writes (video streamming)
-Huge files with random reads/writes (database)
Do you want the best for any one of these, or the best overall?
-Thomas