Linux-Setup Digest #413, Volume #21 Sun, 10 Jun 01 20:13:07 EDT
Contents:
Re: installing alsa 0.9 ("R.V. Gronoff")
Re: Mandrake 8 common /mnt/cdrom locked prob ("still@large")
Re: Hard drive performance w/ hdparm (Mike W.)
Re: console based rpm manager, are there any free utils??? ("Sean McAvoy")
Re: Linux Email Server (Neill Newman)
Re: 2GB File size limitation (Betastar)
Re: Hard drive performance w/ hdparm (Robert Davies)
Re: Linux Email Server (Robert Davies)
Re: Changing root login? (Robert Davies)
Re: swap doesn't unmout properly (Robert Davies)
Re: Hard drive performance w/ hdparm (Michael Perry)
Re: Linux Email Server (Bob Hauck)
Re: /bin/login cannot be removed (John Hasler)
Re: Shift Fx keys in RH Linux (* Tong *)
Re: 2GB File size limitation (Rinaldi J. Montessi)
----------------------------------------------------------------------------
From: "R.V. Gronoff" <[EMAIL PROTECTED]>
Subject: Re: installing alsa 0.9
Date: Mon, 11 Jun 2001 00:18:02 +0200
"croc" <[EMAIL PROTECTED]> a �crit dans le message news:
[EMAIL PROTECTED]
> ok I think I have correctly installed the drivers and libs but when I go
> to the utils, I do ./configure and then 'make' and I get loads of
> 'undefined references' .... any clues anyone?
>
Does it stop the compiling process ? If not, then everything's fine: your
compiled driver will only contain the "material" appropriate to your very
system.
Perhaps this can help:
http://www.linuxnewbie.org/nhf/intel/hardware/yamaha_alsa.html
------------------------------
From: "still@large" <[EMAIL PROTECTED]>
Subject: Re: Mandrake 8 common /mnt/cdrom locked prob
Date: Sun, 10 Jun 2001 22:32:13 GMT
On Sun, 10 Jun 2001 15:44:03 GMT, "still@large" <[EMAIL PROTECTED]>
wrote:
>On Mon, 04 Jun 2001 23:10:59 GMT, sleepy <[EMAIL PROTECTED]>
>wrote:
>
>>Ok. Here's 64 dollar question. There is a permission problem with my
>>cdroms. In Mandrake 8.0, the cdroms are supermounted (automount) for all
>>practical purposes. I get little green icons telling me that they are
>>mounted but I can't access them (GUI or command line). In the GUI file
>>browser Konquerer there are locks on my cdrom(IDE Plexwriter) and
>>cdrom2(IDE Kenwood 42x). They both worked great until I tried to do some
>>scsi emulation b.s. (I just call it that because it's been problematic
>>lately) . That didn't work and I restored my old fstab (to get rid of the
>>scd0 device link. Since then I can't access them. Here's where it gets
>>weird. I created new folders in /mnt called plexcdrom and kenwood and
>>changed my fstab to reflect that. Upon reboot, the cdrom and cdrom2
>>folders were unlocked, but as you could imagine due to fstab, my cdrom's
>>were not there. They were linked to /mnt/plexcdrom and /mnt/kenwood. Yup,
>>you guessed it, there were now locks on those. I am left to believe that
>>the system does not realize I am root and is locking me out of those
>>devices for some reason. If anyone can shed light on this I'll be your
>>best friend for 2 or 3 days. I have thoroughly searched the newsgroups for
>>an answer. Many people have this problem, but no fixes. Below is my fstab
>>FYI
>>
>>fstab
>>
>>/dev/hda6 / ext2 defaults 1 1
>>none /dev/pts devpts mode=0620 0 0
>>/mnt/cdrom /mnt/plexcdrom supermount fs=iso9660,dev=/dev/cdrom 0 0
>>/mnt/cdrom2 /mnt/kenwood supermount fs=iso9660,dev=/dev/cdrom2 0 0
>>/mnt/floppy /mnt/floppy supermount fs=vfat,dev=/dev/fd0 0 0
>>/dev/hdb1 /mnt/windows vfat
>>user,exec,umask=0,codepage=850,iocharset=iso8859-1 0 0
>>/dev/hdb6 /mnt/extra ext2 defaults 1 2
>>none /proc proc defaults 0 0
>>/dev/hda5 swap swap defaults 0 0
>
>I can't help you with your problem exactly, but I can confirm that
>Mandrake 8.0's mount behavior is extremely flaky. I just did this
>install last night, so I haven't had much time to fight with mine. At
>the moment, however, if I mount my cdrom manually (with a full command
>line, not using fstab) it mounts fine, but if I use fstab (which has
>the same devices, options, etc as my command line) it instead mounts
>my floppy drive to the cdrom mount point. Explain that one to me!
>
>I agree with another post that this is very likely due to the
>supermount bs, and I plan on attacking that on my next go-round.
turns out it was the supermount option for me. removing it
immediately fixed everything. if you still have that option included,
you should try removing it.
------------------------------
From: [EMAIL PROTECTED] (Mike W.)
Subject: Re: Hard drive performance w/ hdparm
Date: 10 Jun 2001 15:39:02 -0700
Thanks Mike for the info and links. In other searching I had
found a kernel parameter ("ide2=ata66") that sets up Ultra66
and gives me 25 MB/s (which is better than what I had: 16 MB/s).
I looked at your link and played with hdparm some more, thinking
that the multcount ought to speed things up (32-bit bus access
didn't do anything, prob. b/c the cable is 16-bit no matter
what). But I can't get past 25 MB/s, using UDMA mode 4 (-X68 or
"linux ide2=ata66" at boot time). The drive ought to be doing
40 MB/s on the outer cylinders.
What kernel work did you end up doing to 2.4.5? Maybe it's time
to go look at the source code...sigh. Well, maybe there are
a few more tweaks I can try...
More info: my controller is a CMD 648U, it's on the motherboard,
an Asus CUBX board. My disk drive is a 20GB IBM Deskstar 60GXP 7200
rpm drive (IC35L020AVER07).
-Mike
[EMAIL PROTECTED] (Michael Perry) wrote in message
news:<[EMAIL PROTECTED]>...
> On 8 Jun 2001 11:34:44 -0700, Mike W. <[EMAIL PROTECTED]> wrote:
> > I have Red Hat 7.1 on a ATA/100 IBM drive connected through a
> > CMD Ultra66 interface. The drive is the slave on the first
> > Ultra66 channel (hdf on ide2).
> >
> > But:
> > /sbin/hdparm -tT /dev/hdf
> >
> > reports 16 MB/s, that's 16 megabytes per second, on a drive
> > that can do 40 MB/s sustained read, and through an interface
> > that can do 32 MB/s.
> >
> > What's wrong? On Windows I have another drive running at its
> > rated speed of 29 MB/s (Windows can exploit Ultra66).
>
> You probably have to do a little tuning for hdparm to begin to do its stuff.
> Here is what mine does with some hdparm variables in a start up script:
>
> mperry:/etc/X11# hdparm -Tt /dev/hda
>
> /dev/hda:
> Timing buffer-cache reads: 128 MB in 1.20 seconds =106.67 MB/sec
> Timing buffered disk reads: 64 MB in 2.83 seconds = 22.61 MB/sec
------------------------------
From: "Sean McAvoy" <[EMAIL PROTECTED]>
Subject: Re: console based rpm manager, are there any free utils???
Crossposted-To: linux.redhat.rpm,comp.os.linux.redhat,comp.os.linux,linux.redhat
Reply-To: [EMAIL PROTECTED]
Date: Sun, 10 Jun 2001 22:42:57 GMT
Hello,
rpm will do the trick, but if you want an actual console interface, try
Purp. It's ncurses based.
http://www.lysator.liu.se/purp/
In article <[EMAIL PROTECTED]>, "inetquestion"
<[EMAIL PROTECTED]> wrote:
> I remember using a gui based gnome rpm update utility to keep my system
> updated. Now I want to do the same thing to other machines remotely.
> Redhat has the up2date tool, but that is a pay service. Is there a
> console based rpm utility that can be used to keep a system updated
> which is free?
>
> Thanks,
>
> mailto:[EMAIL PROTECTED]
>
------------------------------
From: Neill Newman <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc,linux.redhat.install,linux.redhat.misc
Subject: Re: Linux Email Server
Date: Mon, 11 Jun 2001 00:06:58 +0100
I second that, we use Exim for SMTP, and Courier for POP3(s) and
IMAP(s).
This supports virtual domains nicely...
HTH
Neill
Chiefy wrote:
>
> 10 Jun 2001 19:32 UTC, Mike typed:
> > Looking for a good Linux Email server applicatoin. Anybody got a good
> > suggestion?
>
> Exim is popular with ISP's here in the UK.
>
> --
> Chiefy. <[EMAIL PROTECTED]>
--
Open Source Specialists http://www.entora.co.uk/
Tel: +44 (0)701 0723686 Fax: +44 (0)870 3214368
------------------------------
From: [EMAIL PROTECTED] (Betastar)
Crossposted-To:
comp.os.linux.misc,comp.os.linux.x,linux.redhat,linux.redhat.install,linux.redhat.misc,redhat.kernel.general
Subject: Re: 2GB File size limitation
Date: Sun, 10 Jun 2001 23:04:35 GMT
On 10 Jun 2001 14:54:43 -0400, [EMAIL PROTECTED] wrote:
>[EMAIL PROTECTED] (Betastar) writes:
>> That's where I'm running into problems too - DNA sequence databases.
>
>Are you using Oracle by any chance? Oracle can easily be configured
>to store data in multiple datafiles.
No - I am trying to download the databses from NCBI, which take over
2GB in space for some of the files just to download. I can't get them
to the machine at all right now short of downloading to a Windoze
machine, splitting them up, and sending them to the Linux machine
afterwards.
>However, whichever way you look at it it will be a pain. You have my
>sympathy.
Thanks. It is a pain. And it's holding up my research something
awful right now.
------------------------------
From: Robert Davies <[EMAIL PROTECTED]>
Subject: Re: Hard drive performance w/ hdparm
Date: Sun, 10 Jun 2001 22:50:17 +0100
Mike W. wrote:
> I have Red Hat 7.1 on a ATA/100 IBM drive connected through a
> CMD Ultra66 interface. The drive is the slave on the first
> Ultra66 channel (hdf on ide2).
>
> The command:
> /sbin/hdparm -i /dev/hdf
>
> reports the drive is in UDMA mode 2, which is Ultra33. I
> believe this is correct, since Linux doesn't exploit Ultra66
> or Ultra100 yet.
It does, and should do, as the IDE driver in recent version (2.2.18 & 2.4)
tune the drive parameters automatically.
All the advice about using hdparm(8) _was_ true but it should no longer be
necessary, according to Andre Hedrik himself, who took part in a thread on
the linux-abit Mailing list.
Now some chipsets and drives are on black lists, (there are some white
lists to around in the kernel, where default is to play safe due to common
faults).
So if you are using a recent 2.4 kernel, and your drive and chipset
combination do not optimise the drive it is most likely to be as a result
of reported corruption.
There's been a lot about the Via Southbridge lately used on boards like the
Asus A7V, so you're best to check on a Linux Kernel Mailing List archive,
for reports of your drive type, chipset and the word corruption, before you
go tweaking.
BTW The -c 1 or -c 3 stuff is obsolete, it's an old EIDE option, and ATA-2
(33 speed), ATA-66 and ATA-100 are the current standards. Similiarly the
-u 1 option is pointless as you have UDMA transfers.
Hopefully the man page will get updated sometime, maybe ...
Rob
------------------------------
From: Robert Davies <[EMAIL PROTECTED]>
Subject: Re: Linux Email Server
Crossposted-To: comp.os.linux.misc,linux.redhat.install,linux.redhat.misc
Date: Sun, 10 Jun 2001 22:38:34 +0100
Vilmos Soti wrote:
> "Mike" <[EMAIL PROTECTED]> writes:
>
>> Looking for a good Linux Email server applicatoin. Anybody got a good
>> suggestion?
>
> What are your expectations?
>
> The four best known are sendmail, exim, postfix, and qmail. The sites for
> them are www.name_of_server.org
He might well mean a POP3 or IMAP server rather than an MTA, in Win world
it
tends to all be done by Exchange. In Linux you get to mix and match, and
there are varying daemons, suiting different niches, depending on
simplicity,
power, security tradeoffs.
So Mike you have a choice, best decide what protocols you wish to serve,
then
type them in at google.com/linux and see what it comes up with.
Rob
------------------------------
From: Robert Davies <[EMAIL PROTECTED]>
Subject: Re: Changing root login?
Date: Sun, 10 Jun 2001 23:25:22 +0100
David Anderson wrote:
> Hello
>
> I would like to know if it is possible to change the login name for the
> root user. Say for example that I want to call him 'admin' instead root,
> how do I go about it? Is it simply a matter of changing the name in
> /etc/passwd and /etc/shadow, or is there more that has to be done?
Better just to copy the root field, and call it admin. Then you can login
as admin if you want to be perverse.
I should not change the 'root' entry, as some programs you install might
use the user name root for some things, rather than the UID. Similarly if
something checks for root ownership in somewhat broken ways, 'admin' would
cause it to worry.
Best get used to root, or you'll find something don't like it being changed.
Rob
------------------------------
From: Robert Davies <[EMAIL PROTECTED]>
Subject: Re: swap doesn't unmout properly
Date: Sun, 10 Jun 2001 23:27:59 +0100
ken pile wrote:
> Some advice please ladies and gentlemen....
>
> My SuSE 7.0 machine sometimes gets stuck when shutting down at the point
> where the swapfile is "unmounted". Unfortunately this is before the
> system partitions have unmounted, so on the next boot all my partitons
> have to be scanned, which is tedious and no doubt unhealthy. Can anyone
> help me work out why this is happening, and is it safe/possible to get
> the system unmounted before the swap partition.
A recent thread on the LKML suggested you should be safe enough just
commenting out 'swapoff', as it's a bit pointless really, unless you are
using swap files, rather than partitions.
Rob
------------------------------
From: [EMAIL PROTECTED] (Michael Perry)
Subject: Re: Hard drive performance w/ hdparm
Date: Sun, 10 Jun 2001 23:37:32 -0000
Reply-To: [EMAIL PROTECTED]
On 10 Jun 2001 15:39:02 -0700, Mike W. <[EMAIL PROTECTED]> wrote:
> Thanks Mike for the info and links. In other searching I had
> found a kernel parameter ("ide2=ata66") that sets up Ultra66
> and gives me 25 MB/s (which is better than what I had: 16 MB/s).
>
> I looked at your link and played with hdparm some more, thinking
> that the multcount ought to speed things up (32-bit bus access
> didn't do anything, prob. b/c the cable is 16-bit no matter
> what). But I can't get past 25 MB/s, using UDMA mode 4 (-X68 or
> "linux ide2=ata66" at boot time). The drive ought to be doing
> 40 MB/s on the outer cylinders.
>
> What kernel work did you end up doing to 2.4.5? Maybe it's time
> to go look at the source code...sigh. Well, maybe there are
> a few more tweaks I can try...
>
> More info: my controller is a CMD 648U, it's on the motherboard,
> an Asus CUBX board. My disk drive is a 20GB IBM Deskstar 60GXP 7200
> rpm drive (IC35L020AVER07).
>
> -Mike
>
Hi-
Well, not very much and some of it was trying to get nvidia's agp support
working. I enabled mtrr support which seemed to help things quite a bit at
a regular level and I got the nvidia agp loading. I also enabled regular
dma support in the 2.4.5 kernel. I just checked to see what I got before my
puny drives get their hdparm assignments and it was less than 2mb/s. Yechh.
I also enabled busmastering dma in the kernel which is in the ide config
block in the 2.4 kernel series. I'm not using terribly good drives but the
increase I see is pretty significant. I get up around 25 - 30 depending on
the frequency of running the hdparm tests. There may have been other kernel
changes I made also. I'll look at the .config later tonite.
Prior to this, at 2mb, I got serious keyboard buffer and file copy problems
in xterms and the console. The console would get sluggish, xterms would
really get slow and the mouse would jerk all around. The results are pretty
much like what the oreilly article said.
I think the primary difference I am seeing is with the hdparm stuff, with
nvidia's agp stuff working, and my athlon thunderbird does much better with
a 2.4 kernel I think.
I actually have two athlons that I use. This one has dual maxtor drives;
the other has a western digital drive. I am getting very good acceleration
in udma4 on that system now after playing with the drive and hdparm for
awhile.
I managed to find yet another hdparm article off of google. I'll replay the
ultraDMA mode stuff here:
o Ultra DMA Mode
o 0 15mb/sec
o 1 25mb/sec
o 2 -X66 33mb/sec udma33
o 3 50mb/sec
o 4 -X68 66mb udma66
o 5 -X69 100mb udma100
You can also use multiword dma mode like:
o Multiword DMA Mode
o 0 -X32 4.25mb
o 1 -X33 13.33mb
o 2 -X34 16.6mb
I think if I bought better drives, with some of the stuff I have managed to
learn about hdparm and the kernel, I could get very good performance. Dave
Uhring reported some nice performance gains on his system also. He uses IBM
drives as well.
Did you see his post? Its in this thread.
--
Michael Perry
[EMAIL PROTECTED]
====================
------------------------------
From: [EMAIL PROTECTED] (Bob Hauck)
Crossposted-To: comp.os.linux.misc,linux.redhat.install,linux.redhat.misc
Subject: Re: Linux Email Server
Reply-To: bobh = haucks dot org
Date: Sun, 10 Jun 2001 23:44:02 GMT
On Sun, 10 Jun 2001 15:05:31 -0700, Mike <[EMAIL PROTECTED]> wrote:
> > The four best known are sendmail, exim, postfix, and qmail. The sites for
> > them are www.name_of_server.org
> I am fairly new to Linux and need an Email Server that can handle virtual
> domains similiar to Sealtle Labs SL Mail for NT and Win 2k.
All of the ones he mentioned can do virtual domains. You can combine
any of them with an imap and/or pop3 server (uw-imap, qpopper, cyrus,
etc) to get the full solution you are asking for.
Unix mail systems tend to give you a lot of options for mix-and-match.
--
-| Bob Hauck
-| To Whom You Are Speaking
-| http://www.haucks.org/
------------------------------
From: John Hasler <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.misc
Subject: Re: /bin/login cannot be removed
Date: Sun, 10 Jun 2001 22:17:25 GMT
Peter writes:
> His distro surely provides him with tools (tripwire? rpm? - does nayone
> kow what the debian equivalent is, if there is one?) that will check his
> installation for him.
The Debian equivalent of tripwire is, of course, tripwire. Also of
interest may be debsums, which handles md5sums for installed packages.
--
John Hasler
[EMAIL PROTECTED] (John Hasler)
Dancing Horse Hill
Elmwood, WI
------------------------------
From: * Tong * <[EMAIL PROTECTED]>
Crossposted-To: comp.emacs
Subject: Re: Shift Fx keys in RH Linux
Date: 10 Jun 2001 20:53:53 -0300
"Mark A. Flacy" <[EMAIL PROTECTED]> writes:
> Is this under X-windows?
yes.
> If so, does your window manager intercept the Shift-Fx keys?
> What does C-h c S-F1 tell you?
no. It says F13, and S-F2 is F14, so forth...
As I said, if I start my X server using VNC server, and connect
using VNC viewer, everything works fine, for the same window manager
(Fvwm2) and same ... everything.
--
Tong (remove underscore(s) to reply)
*niX Power Tools Project: http://xpt.sourceforge.net/
- All free contribution & collection
------------------------------
Crossposted-To:
comp.os.linux.misc,comp.os.linux.x,linux.redhat,linux.redhat.install,linux.redhat.misc,redhat.kernel.general
From: [EMAIL PROTECTED] (Rinaldi J. Montessi)
Subject: Re: 2GB File size limitation
Date: Sun, 10 Jun 2001 23:54:49 GMT
Betastar wrote:
> On 10 Jun 2001 14:54:43 -0400, [EMAIL PROTECTED] wrote:
>
>>[EMAIL PROTECTED] (Betastar) writes:
>>> That's where I'm running into problems too - DNA sequence databases.
>>
>>Are you using Oracle by any chance? Oracle can easily be configured
>>to store data in multiple datafiles.
>
> No - I am trying to download the databses from NCBI, which take over
> 2GB in space for some of the files just to download. I can't get them
> to the machine at all right now short of downloading to a Windoze
> machine, splitting them up, and sending them to the Linux machine
> afterwards.
>
>>However, whichever way you look at it it will be a pain. You have my
>>sympathy.
>
> Thanks. It is a pain. And it's holding up my research something
> awful right now.
I'm posting something I read in this group a few days ago. It may bear
on your problem. If not please post back with why it doesn't help - or
why it does?
<Quote>
0. The limitation is twofold and is not filesystem dependent.
1. The limitation arises from the decision to store some things in
processor-native data types in kernel 2.2 and 2.0.
2. The x86 architecture uses a 32-bit int for many things.
3. Therefore, on older kernels on the x86 architecture, file size is
limited to the values you can store in a 32-bit signed int.
4. Older glibc and all the programs linked against it inherit this
32-bit signed int limitation, leading to the 2G problem. It never was a
problem on 64-bit architectures like Alpha and Sparc.
So, what can you do? Upgrade the kernel to 2.4.5, recompile your glibc
against the new kernel, and recompile the applications you use against
this new glibc. Then everything will use 64 bits for file sizes and
file position offsets, and you can have 2T files.
The fact that you're using RH 7.1 and having this problem surprises me.
We have a server running a stock RedHat 7.0 install, and just yesterday,
I created a 5G file on its disk. No problems at all. Check RedHat's
website, search for "large file", see what you find? It would not
surprise me if there were a few RPMs you could download for large file
support.
(BTW, Jonas, ReiserFS has its own filesize limitation, but it's 4G.
ext2's limit is 2T. Next time, make sure you know what you're talking
about before posting, OK?)
--
Matt G|There is no Darkness in Eternity/But only Light too dim for us to see
Brainbench MVP for Linux Admin / Outside of a dog, a book is a man's best
http://www.brainbench.com / friend. Inside of a dog, it's too dark
=============================/ to read. ==Groucho Marx
</Quote>
Rinaldi
--
We all know Linux is great...it does infinite loops in 5 seconds.
--Linus Torvalds
------------------------------
** FOR YOUR REFERENCE **
The service address, to which questions about the list itself and requests
to be added to or deleted from it should be directed, is:
Internet: [EMAIL PROTECTED]
You can send mail to the entire list by posting to comp.os.linux.setup.
Linux may be obtained via one of these FTP sites:
ftp.funet.fi pub/Linux
tsx-11.mit.edu pub/linux
sunsite.unc.edu pub/Linux
End of Linux-Setup Digest
******************************
