Linux-Misc Digest #513, Volume #27 Mon, 2 Apr 01 11:13:02 EDT
Contents:
how can I avoid a reboot? (bruce)
Re: Tar with multi-volumes ("Eric")
nfssvc: Function not implemented RH-6.2 (Kevin Taylor)
Re: Tar with multi-volumes (Olivier Colliot)
Kernel parameters ("Bastiaan Schaap")
Re: Tar with multi-volumes ("Eric")
Re: nfssvc: Function not implemented RH-6.2 ([EMAIL PROTECTED])
Re: Kernel parameters ("Eric")
Re: tar backup via NFS excluding NFS mount point (Lee Allen)
Re: system.map ("Peter T. Breuer")
Re: ftp Login incorrect ("Peter T. Breuer")
Re: Kernel parameters ("Bastiaan Schaap")
LILO+w98 ("Mait Lang")
Re: system.map ("Eric")
Re: Kernel parameters ("Eric")
Re: LILO+w98 ("Eric")
Re: nfssvc: Function not implemented RH-6.2 (Kevin Taylor)
Re: mem and swap problem (Gabor Takacs)
Re: nfssvc: Function not implemented RH-6.2 ([EMAIL PROTECTED])
Re: how can I avoid a reboot? ("Jason C. Hill")
Re: LILO+w98 ("Mait Lang")
Re: ISO image install problem... ("Tauno Voipio")
----------------------------------------------------------------------------
From: bruce <[EMAIL PROTECTED]>
Subject: how can I avoid a reboot?
Date: Mon, 02 Apr 2001 08:19:08 -0400
running rh 6.1, kernel 2.2.16
I had a problem w/ squid the other day. It died and I could restart it.
It stated another process was listening on one of the tcp ports it uses.
Netstat did show this port in use. However none of the processes
associated w/ squid were running. The only way I could clear it was to
reboot.
Is there another way to clear a specific tcp port with having to reboot
the whole box? TIA.
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: Tar with multi-volumes
Date: Mon, 2 Apr 2001 14:24:03 +0200
> 1) I would like to create a tar archive divided in several files
> (file1.tar, file2.tar ...), each file being less than 650Mb big.
> I have tried tar -L 650000 -M but it doesn't work (it seems to only
> handle multi-tapes not multi-files)
> Does anybody know how to do this ?
I'm not sure if this can be done.
but you can cut an existing tar file in 650M pieces. (like the -L option
does)
> 2) Is it true that on an ext2 filesystem the maximum size for a file is
> 2Gb ?
definitely not true.
If you run in to that limit, it's an architecture limit, due to your
processor.
There are patches (LFS patch) but you'd need to update your tools too,
in order to fully use larger files. The new 2.4 kernels, can by default
handle larger files.
> What about other fs (like reiserfs or anything else) ?
same for these.
Eric
------------------------------
From: Kevin Taylor <[EMAIL PROTECTED]>
Subject: nfssvc: Function not implemented RH-6.2
Date: Mon, 02 Apr 2001 09:05:11 -0400
I'm using kernel 2.2-17...and after several rebuilds of the kernel I
cannot get NFS to start up. Each time it gives me the same error message
(see subject).
Any hints?
------------------------------
From: Olivier Colliot <[EMAIL PROTECTED]>
Subject: Re: Tar with multi-volumes
Date: Mon, 02 Apr 2001 15:14:12 +0200
>
Thanks for your answer
> I'm not sure if this can be done.
> but you can cut an existing tar file in 650M pieces. (like the -L option
> does)
Should I use 'split' to cut the file ? (which means i have to merge them
before untar, I guess)
Olivier
------------------------------
From: "Bastiaan Schaap" <[EMAIL PROTECTED]>
Subject: Kernel parameters
Date: Mon, 2 Apr 2001 15:43:12 +0200
Hi all,
I'm currently installing Oracle on a linux box with Red Hat 6.2 (kernel
2.2.14). My installation guide says I need to configure a set of kernel
parameters. How/where can I set these parameters, and does it require kernel
recompilation (not that that's a problem though....)?? I actually found one
parameter: SHMMAX, which resides under /proc/sys/kernel as a 'file' called
shmmax. I cannot change it with vi, but I found out that: echo 0x8000000 >
shmmax works in my case. However, when I reboot this setting is lost. What
do I need to do in order to get all parameters into the system correctly,
and permanent?
TIA,
Bastiaan Schaap
Desyde BV
[EMAIL PROTECTED]
Tel 06 - 51998277
Fax 035 - 5430547
http://www.desyde.nl
________________________________
'The Computer made me do it.'
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: Tar with multi-volumes
Date: Mon, 2 Apr 2001 15:42:08 +0200
> > I'm not sure if this can be done.
> > but you can cut an existing tar file in 650M pieces. (like the -L option
> > does)
>
> Should I use 'split' to cut the file ? (which means i have to merge them
> before untar, I guess)
That's the kind of tool I had in mind, yes.
And indeed, before untarring, they have to be merged (with cat eg.) first.
Eric
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: nfssvc: Function not implemented RH-6.2
Date: 2 Apr 2001 13:44:44 GMT
Kevin Taylor <[EMAIL PROTECTED]> wrote:
> I'm using kernel 2.2-17...and after several rebuilds of the kernel I
> cannot get NFS to start up. Each time it gives me the same error message
> (see subject).
> Any hints?
If you are using the Kernel space version of the NFS deamon you need
NFS support inside the kernel. See the NFS-HOWTO.
Davide
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: Kernel parameters
Date: Mon, 2 Apr 2001 15:47:16 +0200
> I'm currently installing Oracle on a linux box with Red Hat 6.2 (kernel
> 2.2.14). My installation guide says I need to configure a set of kernel
> parameters. How/where can I set these parameters, and does it require
kernel
that's probably an option, but not required in your case appearantly.
> recompilation (not that that's a problem though....)?? I actually found
one
> parameter: SHMMAX, which resides under /proc/sys/kernel as a 'file' called
> shmmax. I cannot change it with vi,
that's good, in /proc no real files can be found. The files you see there
are
representations of kernel settings.
> but I found out that: echo 0x8000000 >
> shmmax works in my case. However, when I reboot this setting is lost. What
> do I need to do in order to get all parameters into the system correctly,
> and permanent?
You can put it in your rc.local.
That way it gets run automatically when you boot.
Eric
------------------------------
From: [EMAIL PROTECTED] (Lee Allen)
Subject: Re: tar backup via NFS excluding NFS mount point
Reply-To: [EMAIL PROTECTED]
Date: Mon, 02 Apr 2001 13:48:05 GMT
On 30 Mar 2001 17:37:56 +0200, Stefano Ghirlanda <[EMAIL PROTECTED]>
wrote:
>[EMAIL PROTECTED] (Lee Allen) writes:
>
>> On 29 Mar 2001 17:52:17 GMT, [EMAIL PROTECTED] (Scrumpy) wrote:
>>
>> The simpler way to accomplish your goal is "--one-file-system"
>
>But it's the answer to another question...
>
>He asked: how do I exclude /cdrom, not 'how do I limit operation to
>the / filesystem. If / is only a boot partition and /usr, /home and so
>on are on other partitions the two things are different.
1) He is a newbie and I assumed he did not realize there are other
things he needs to exclude. The first several times I backed up "/" I
inadvertantly included /proc which wasted a huge amount of space and
generated many errors. I thought I would help him anticipate & avoid
those problems.
2) You're right: By the same thinking it is remotely possible he has
other filesystems mounted under "/" which need to be backed up and he
may not be aware of it. In that case they should be specified as
add'l directories to be backed up, or better yet, should be backed up
separately.
-Lee Allen
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: system.map
Date: Mon, 2 Apr 2001 15:35:59 +0200
green <[EMAIL PROTECTED]> wrote:
> I have no clue on what an alternator may be
Then clearly you don't have a dictionary, or can't read the one you
have.
The original point here was that a person who doesn't know what
System.map is for _certainly_ does not need it! It's like saying that
you don't know what a microscope is for.
Peter
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: ftp Login incorrect
Date: Mon, 2 Apr 2001 15:37:47 +0200
Utama Abdulwahid <[EMAIL PROTECTED]> wrote:
> I'm using a RedHat 7 server. Recently I can't ftp in from anywhere with
> any username because of "Login incorrect". Can anyone point me to things
> I should check for or if I need to reinstall anything?
> The last thing I remember was upgrading my PHP and Apache.
Hmm .. and which public bar did you go to?
Sounds like you upgraded wu-ftpd while amnesiac. Or someone hacked it
while you were out for the count.
Peter
------------------------------
From: "Bastiaan Schaap" <[EMAIL PROTECTED]>
Subject: Re: Kernel parameters
Date: Mon, 2 Apr 2001 16:00:28 +0200
Hi Eric,
Thanx! I've got that working! However, I have other parameters/semaphores
that I need to set, which are not shown in /proc/sys/kernel... Can I do an
echo into them anyway? Or is there another place where I have to set these?
--
Bastiaan Schaap
Desyde BV
[EMAIL PROTECTED]
Tel 06 - 51998277
Fax 035 - 5430547
http://www.desyde.nl
________________________________
'Hurhurhur, 2400 baud sucks' - V.bis and Baudhead
"Eric" <[EMAIL PROTECTED]> wrote in message
news:9a9vt4$4k3$[EMAIL PROTECTED]...
> > I'm currently installing Oracle on a linux box with Red Hat 6.2 (kernel
> > 2.2.14). My installation guide says I need to configure a set of kernel
> > parameters. How/where can I set these parameters, and does it require
> kernel
>
> that's probably an option, but not required in your case appearantly.
>
> > recompilation (not that that's a problem though....)?? I actually found
> one
> > parameter: SHMMAX, which resides under /proc/sys/kernel as a 'file'
called
> > shmmax. I cannot change it with vi,
>
> that's good, in /proc no real files can be found. The files you see there
> are
> representations of kernel settings.
>
> > but I found out that: echo 0x8000000 >
> > shmmax works in my case. However, when I reboot this setting is lost.
What
> > do I need to do in order to get all parameters into the system
correctly,
> > and permanent?
>
> You can put it in your rc.local.
> That way it gets run automatically when you boot.
>
> Eric
>
>
------------------------------
From: "Mait Lang" <[EMAIL PROTECTED]>
Subject: LILO+w98
Date: Mon, 2 Apr 2001 16:02:57 +0200
Hi,
I'm having problem with booting w98 at lilo prompt. When i type in w98
(image name) then system just hangs after enter key (or last time i got half
a screen ascii marks) and only reset key helps. No problem to boot linux. I
set w98 as default in lilo. conf and this is the only way now to boot w98
(after timeout without writing image name at lilo prompt it goes up without
problems).
The problem started to appeare randomly at boot and now it is stable. I
reinstalled lilo but no changes:(
any ideas?
Mait
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: system.map
Date: Mon, 2 Apr 2001 16:00:47 +0200
> > I have no clue on what an alternator may be
>
> Then clearly you don't have a dictionary, or can't read the one you
> have.
Hi peter,
I do have one, but just never bothered to look.
I really wasn't that interested in the meaning of the word.
> The original point here was that a person who doesn't know what
> System.map is for _certainly_ does not need it!
Exactly.
Eric
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: Kernel parameters
Date: Mon, 2 Apr 2001 16:07:28 +0200
> Hi Eric,
>
> Thanx! I've got that working! However, I have other parameters/semaphores
> that I need to set, which are not shown in /proc/sys/kernel... Can I do an
> echo into them anyway? Or is there another place where I have to set
these?
That all depends.
Some modules can be loaded with explicit settings (interrupt, io).
If it's not in a module, I doubt it can be done. (Or perhaps through IOCTL
calls)
I really don't know much about the exact thing you want to achieve, so read
the
documentation that's included and follow that advise. If it involves
changing
parameters in the kernel source tree, go ahead. That's one of the advantages
of an open-source OS.
Eric
------------------------------
From: "Eric" <[EMAIL PROTECTED]>
Subject: Re: LILO+w98
Date: Mon, 2 Apr 2001 16:13:18 +0200
> I'm having problem with booting w98 at lilo prompt. When i type in w98
> (image name) then system just hangs after enter key (or last time i got
half
> a screen ascii marks) and only reset key helps. No problem to boot linux.
I
> set w98 as default in lilo. conf and this is the only way now to boot
w98
> (after timeout without writing image name at lilo prompt it goes up
without
> problems).
No way. Then they're not the same.
Did you rerun `/sbin/lilo -v`
What's in your lilo.conf? (show it please)
> The problem started to appeare randomly at boot and now it is stable. I
> reinstalled lilo but no changes:(
What do you mean with reinstall?
just ran /sbin/lilo or did you install a newer version?
Eric
------------------------------
From: Kevin Taylor <[EMAIL PROTECTED]>
Subject: Re: nfssvc: Function not implemented RH-6.2
Date: Mon, 02 Apr 2001 10:19:28 -0400
[EMAIL PROTECTED] wrote:
>
> Kevin Taylor <[EMAIL PROTECTED]> wrote:
> > I'm using kernel 2.2-17...and after several rebuilds of the kernel I
> > cannot get NFS to start up. Each time it gives me the same error message
> > (see subject).
> > Any hints?
>
> If you are using the Kernel space version of the NFS deamon you need
> NFS support inside the kernel. See the NFS-HOWTO.
>
> Davide
I've tried compiling with kernel support and also with just module
support...none of it is working.
------------------------------
From: Gabor Takacs <[EMAIL PROTECTED]>
Crossposted-To:
alt.linux,alt.os.linux,comp.os.linux.setup,linux.misc,linux.redhat,linux.redhat.misc,linux.support.commercial,redhat.config,redhat.general
Subject: Re: mem and swap problem
Date: Mon, 02 Apr 2001 10:15:39 -0400
The exact value is 127+change and it does work. It was provided as a solution to
this problem. If need the exact value try the external linux.redhat.misc
newsgroup, that is where I got my info. This solution was applied successfully
to v6.x of RedHat and Mandrake.
Taavi Hein wrote:
> "Gabor Takacs" <[EMAIL PROTECTED]> wrote in message
> news:[EMAIL PROTECTED]...
> : I read something about this on a linux.redhat newsgroup. Apparently some
> : portion of the RAM is used for something else (store a copy of BIOS I
> : think). Because of this, you don't have the entire 128M. I don't
> : remember the exact number.
>
> Still, suppose, you specified a number in boot options, Linux is gonna
> think, that's how much you got, right? Then, it uses a few kB for smth (why
> is a copy of bios necessary for Linux?), if you specified a few kB less, it
> would use the few kB for smth, but you'd have twice less space, than when
> specifying the correct value [if I'm not making any sense ... just disregard
> the letter]
>
> --
> Taavi Hein - [EMAIL PROTECTED]
> Registered Linux user #209546
> Registered Linux machine #97395
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: nfssvc: Function not implemented RH-6.2
Date: 2 Apr 2001 14:42:28 GMT
Kevin Taylor <[EMAIL PROTECTED]> wrote:
> I've tried compiling with kernel support and also with just module
> support...none of it is working.
Well, I had problems myself before have this working, this was my
path: first I'm using 2.4.0, I build with NFS server/client support
in the kernel, then I had to update the NFS deamon software
(I downloaded the new version of the deamons and recompiled), then
I had to define the new /etc/exports and setup some permission
under /etc/hosts.allow and /etc/hosts.deny (I discovered that if
you don't have the correct permission for portmap into the
hosts.allow the system wont start.
After that, the nfsd was working fine on my system. On another
machine with kernel 2.2.17, the NFS was working without doing
nothing (very upsetting...).
I suggest you revise the NFS-HOWTO and check for "troubleshooting"
in the last part of the howto...
Davide
------------------------------
From: "Jason C. Hill" <[EMAIL PROTECTED]>
Subject: Re: how can I avoid a reboot?
Date: Mon, 2 Apr 2001 10:44:34 -0400
Usually if the process behind the port that it's listening to died, then
give it a few minutes (shouldn't have to wait more than 5) and the port will
eventually die out on you (and thus can be re-used). If you're in a big
hurry to get it back up, then go ahead and reboot.
As for a tool that will kill off a port so that it can be reused, no, there
isn't any such tool to my knowledge.
-J
"bruce" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> running rh 6.1, kernel 2.2.16
>
> I had a problem w/ squid the other day. It died and I could restart it.
> It stated another process was listening on one of the tcp ports it uses.
> Netstat did show this port in use. However none of the processes
> associated w/ squid were running. The only way I could clear it was to
> reboot.
>
> Is there another way to clear a specific tcp port with having to reboot
> the whole box? TIA.
------------------------------
From: "Mait Lang" <[EMAIL PROTECTED]>
Subject: Re: LILO+w98
Date: Mon, 2 Apr 2001 16:40:59 +0200
> No way. Then they're not the same.
Looking with the eye they were but i discovered with vi an enter after the
last row (label = w98) and one empty row after that. But this shall not be a
reason for random w98 boot failure at lilo prompt.
> Did you rerun `/sbin/lilo -v`
no
> What's in your lilo.conf? (show it please)
boot = /dev/hda
timeout = 100
linear
prompt
default = w98
vga = normal
root = /dev/hda4
read-only
map=/boot/map
install=/boot/boot.b
image = /boot/vmlinuz-2.2.14-5.0
label = linux
other = /dev/hda1
label = w98
Im using lilo-0.21-15
>
> > The problem started to appeare randomly at boot and now it is stable. I
> > reinstalled lilo but no changes:(
>
> What do you mean with reinstall?
> just ran /sbin/lilo or did you install a newer version?
Just ran lilo.
Mait.
------------------------------
From: "Tauno Voipio" <[EMAIL PROTECTED]>
Crossposted-To: comp.os.linux.setup,linux.redhat.misc
Subject: Re: ISO image install problem...
Date: Mon, 02 Apr 2001 14:49:14 GMT
"Rod Smith" <[EMAIL PROTECTED]> wrote in message
news:[EMAIL PROTECTED]...
> In article <[EMAIL PROTECTED]>,
> [EMAIL PROTECTED] (steve) writes:
> >>|
> >>| You should have the ISO image back on the hard disk of the burner
machine
> >>| and make a new CD with 'Make CD from image' option in the burner
program.
> >
> > What I usually do is mount the image then burn to cd. Works everytime.
>
> That can easily lose information, particularly when dealing with
> cross-OS burning. For instance, it's not clear if the original poster
> burned from within Linux. If the original poster was using Windows,
> then mounting the image is iffy (Windows itself doesn't support it, but
> I believe there are third-party utilities that do). Assuming the image
> is mounted in Windows, you'd lose stuff like Rock Ridge owner and
> permissions information, some of which may be important for a Linux
> installation CD. When doing this in Linux, you might lose the bootable
> nature of the original, unless you take special measures to preserve
> it. Likewise, if the CD image supported both ISO-9660 and HFS, you'd
> lose the HFS side (unless you mounted it as HFS, in which case you might
> lose Rock Ridge information).
>
> All in all, if you download a CD image file, it's best to burn it as an
> image file, not mount it and try to re-parse it. Doing the latter just
> introduces possibilities for error, both subtle and not-so-subtle.
>
Yes. It is already shown too many times that the Windows burners mess up the
file name cases and the installation is lost. Gatesware does not honour the
Rock Ridge extensions.
Tauno Voipio
tauno voipio @ iki fi
------------------------------
** 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.misc.
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-Misc Digest
******************************
