Linux-Misc Digest #520, Volume #27 Tue, 3 Apr 01 12:13:02 EDT
Contents:
Re: What's a good smtp server? (Rod Smith)
loopback device (Alexander Elsenaar)
Re: startx gives problems (aflinsch)
Re: Secure File deletion ("Peter T. Breuer")
Re: Deleting core dumps ("Peter T. Breuer")
MDI SCSI Express Library (cmendlik)
Re: manual fsck and root passwd (aflinsch)
Re: manual fsck and root passwd (aflinsch)
Re: manual fsck and root passwd (aflinsch)
Re: Listening to more than one IP on one ethernet card (Daren Russell)
Re: Win2K messed after RH 6.1 installation ("Tauno Voipio")
Re: difference between ext2fs and reiserfs ([EMAIL PROTECTED])
Re: loopback device (Kent Robotti)
Daylight savings time (Bill Grzanich)
Re: difference between ext2fs and raiserfs ("Peter T. Breuer")
Re: loopback device ("Peter T. Breuer")
Re: loopback device (Alexander Elsenaar)
----------------------------------------------------------------------------
From: [EMAIL PROTECTED] (Rod Smith)
Subject: Re: What's a good smtp server?
Date: Tue, 03 Apr 2001 14:46:54 -0000
Reply-To: [EMAIL PROTECTED]
In article <[EMAIL PROTECTED]>,
[EMAIL PROTECTED] (Dowe Keller) writes:
> Hello all, I am looking for a small, easy to set up smtp server for
> Linux. I recently installed Mandrake 7.2 and it didn't install one.
You're almost certainly mistaken. Mandrake ships with both sendmail and
Postfix, but it uses Postfix by default. (At least, whenever I've tried
it; I've seen posts from others who say it uses sendmail.) My Mandrake
7.2 installation configured Postfix to run at boot time. Try this:
ps ax | grep postfix
You should see a line listing /usr/lib/postfix/master as running. You
can also try:
telnet localhost 25
If you get a banner announcing the machine name and "ESMTP Postfix",
then it's running. (Type "quit" to end the telnet session to the mail
server.)
If you're NOT seeing evidence of Postfix running, type this:
rpm -q postfix
If you get a message saying that Postfix isn't installed, install it
from the RPMs on your Mandrake install CD. If the system replies with
the full package name, then it's installed but isn't running for some
reason. Probably the startup files in /etc/rc.d/rc?.d are set to kill it
rather than start it -- that is, the relevant Postfix startup script's
filename starts with K rather than S. You can fix this with a tool like
ntsysv or tksysv, or manually by renaming the script appropriately (it
should be S80postfix) in each runlevel directory in which you want it to
run (probably 3 and 5, but *NOT* 1, 2, or 6).
--
Rod Smith, [EMAIL PROTECTED]
http://www.rodsbooks.com
Author of books on Linux & multi-OS configuration
------------------------------
From: Alexander Elsenaar <[EMAIL PROTECTED]>
Subject: loopback device
Date: Tue, 03 Apr 2001 14:41:40 GMT
Has anyone problems with the loopback device with a 2.4.2 kernel?
I recompiled the kernel without any problems and now I am unable to
mount/access a loopback filesystem. The system completely deadlocks! This
seemed to be a problem with older kernels but I cannot find any
registration a this problem with the 2.4.2 kernel.
Anybody?
Alexander Elsenaar
------------------------------
From: aflinsch <[EMAIL PROTECTED]>
Subject: Re: startx gives problems
Date: Tue, 03 Apr 2001 10:40:26 -0400
Eric wrote:
>
> > > Last fall I had the same problem with RH 6.2 I don't remember the
> > > exact fix, but I thrashed around looking for xfs lock files (while xfs
> > > was not running), restarted xfs ("/etc/rc.d/init.d/xfs start" [1]) and
> > > xfs was up and running, w/o any problems since. There was definitely
> > > no serious work involved.
> >
> > Thanks for your suggestion. I exactly did what you tried to explain here.
> > First I stopped the xfs service and then removed the xfs lock files under
> > the /var/lock/subsys directory, then started the xfs service and then gave
> > the startx command. But no joy, it is giving the same error and when I ask
> > for the status of xfs it is once again displaying the same message 'xfs
> dead
> > by subsys locked'
> > Here is the sequence of commands that I used.
> >
> > >service xfs stop
> > >rm -fr /var/lock/subsys/xfs
> > >service xfs start
> > >startx
> > >service xfs status
> >
>
> If you didn't change anything prior to this, run df.
> I suppose /tmp or /var to be filled 100%
could also be /home that got full, and noplace to create the
.xsession-errors file -- That's what happened to me.
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Secure File deletion
Date: Tue, 3 Apr 2001 16:41:46 +0200
Jean-David Beyer <[EMAIL PROTECTED]> wrote:
> "Peter T. Breuer" wrote:
>>
>> Hugh Potter <[EMAIL PROTECTED]> wrote:
>> > Im a recent convert to Linux. Ld like to be able to securely delete swap and
>> > other temp files in Linux Mandrake. In Windows I used scorch for the swap file
>> > and eraser for the rest. Any suggestions for linux.
>>
>> man dd. Overwrite it with zeros (then delete it, if appropriate).
>>
> In USA, the department of defense's policy for secure deletion of
> files involves removing the hard drive in question from the machine,
> and shredding the whole thing (a paper shredder will not do it). It
> all depends on just how secure you wish to be.
uh, yep. One could always give it to the DoD and ask them to fly it to
china ... ;.)
Peter
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: Deleting core dumps
Date: Tue, 3 Apr 2001 16:43:07 +0200
Jean-David Beyer <[EMAIL PROTECTED]> wrote:
> but I have never seen that happen. It is possible for a process to
> voluntarily write out a core file at anytime (or, at least, it used to
> be possible in the old UNIX days), but I cannot imagine why an xterm
> would do that.
Possibly to let you do a fast startup the next time ;.).
Peter
------------------------------
From: cmendlik <[EMAIL PROTECTED]>
Subject: MDI SCSI Express Library
Date: Tue, 03 Apr 2001 15:02:56 GMT
I have an MDI SCSI Express Library 5-41DS, Model # C1115H. I would like to
hook this up to my Linux box, does anybody know where I could find drivers
for this device? I am running Red Hat 7.0 Intel.
Thanks!
--
===========
Christopher G. Mendlik
[EMAIL PROTECTED]
------------------------------
From: aflinsch <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux.mandrake
Subject: Re: manual fsck and root passwd
Date: Tue, 03 Apr 2001 10:45:32 -0400
Eric wrote:
>
> > The crahes comes from insmod my own kernel module where the disc buffers
> > are not flushed. This partition is damaged. (To try out simple compile and
> > insmod
> > a kernel module with panic(), this should have the same behavior). By
> booting
>
> well, thanks, but I think I'll pass. :-)
>
> > using a SuSE CD I can use fsck for my rh/mandrake dist 8)
>
> having the rootfs damaged could do strange things, I suppose.
>
> > > > What is ging on here and how can I prevent this ? SuSE is working for
> > > > in those cases !
> > >
> > > Just boot from a resue system and run fsck from there.
> > > I have no idea why you're denied root login.
> >
> > This is the interesting point. It could bee that the other fs are not
> mounted
> > yet,
> > therefore the login process does see only a shadowed passwd file ? My
> > partitions
> > /mountpoints are
> > /, /boot, /var, /tmp, /home,/usr./usr/src
> > Could this possible?
>
> no, all the stuff needed to login is at your / FS (in /etc and /sbin)
/etc is under /, if / is damaged and the part that is damaged happens
to be where /etc is, then the damage might be enough to prevent login.
------------------------------
From: aflinsch <[EMAIL PROTECTED]>
Subject: Re: manual fsck and root passwd
Date: Tue, 03 Apr 2001 10:47:20 -0400
Eric wrote:
>
> > The crahes comes from insmod my own kernel module where the disc buffers
> > are not flushed. This partition is damaged. (To try out simple compile and
> > insmod
> > a kernel module with panic(), this should have the same behavior). By
> booting
>
> well, thanks, but I think I'll pass. :-)
>
> > using a SuSE CD I can use fsck for my rh/mandrake dist 8)
>
> having the rootfs damaged could do strange things, I suppose.
>
> > > > What is ging on here and how can I prevent this ? SuSE is working for
> > > > in those cases !
> > >
> > > Just boot from a resue system and run fsck from there.
> > > I have no idea why you're denied root login.
> >
> > This is the interesting point. It could bee that the other fs are not
> mounted
> > yet,
> > therefore the login process does see only a shadowed passwd file ? My
> > partitions
> > /mountpoints are
> > /, /boot, /var, /tmp, /home,/usr./usr/src
> > Could this possible?
>
> no, all the stuff needed to login is at your / FS (in /etc and /sbin)
/etc is under /, if / is damaged and the part that is damaged happens
to be where /etc is, then the damage might be enough to prevent login.
------------------------------
From: aflinsch <[EMAIL PROTECTED]>
Crossposted-To: alt.os.linux.mandrake
Subject: Re: manual fsck and root passwd
Date: Tue, 03 Apr 2001 10:46:25 -0400
Eric wrote:
>
> > The crahes comes from insmod my own kernel module where the disc buffers
> > are not flushed. This partition is damaged. (To try out simple compile and
> > insmod
> > a kernel module with panic(), this should have the same behavior). By
> booting
>
> well, thanks, but I think I'll pass. :-)
>
> > using a SuSE CD I can use fsck for my rh/mandrake dist 8)
>
> having the rootfs damaged could do strange things, I suppose.
>
> > > > What is ging on here and how can I prevent this ? SuSE is working for
> > > > in those cases !
> > >
> > > Just boot from a resue system and run fsck from there.
> > > I have no idea why you're denied root login.
> >
> > This is the interesting point. It could bee that the other fs are not
> mounted
> > yet,
> > therefore the login process does see only a shadowed passwd file ? My
> > partitions
> > /mountpoints are
> > /, /boot, /var, /tmp, /home,/usr./usr/src
> > Could this possible?
>
> no, all the stuff needed to login is at your / FS (in /etc and /sbin)
/etc is under /, if / is damaged and the part that is damaged happens
to be where /etc is, then the damage might be enough to prevent login.
------------------------------
From: [EMAIL PROTECTED] (Daren Russell)
Subject: Re: Listening to more than one IP on one ethernet card
Date: Tue, 03 Apr 2001 15:22:08 GMT
On Mon, 02 Apr 2001 17:03:00 -0600, Fil Sapienza <[EMAIL PROTECTED]>
wrote:
>Greetings. I would like to set up a virtual host on my Linux box. Is it
>true that
>I can tell one ethernet card to listen to more than one IP address?
>How do I set up Linux to recognize multiple IP addresses on one card?
>
>Thank you.
>
>Filipp Sapienza
>[EMAIL PROTECTED]
>
Hi,
You need to make sure you have IP Aliasing enabled in your kernel.
Then you just do a:
ifconfig eth0:x the.ip.add.ress netamask the.net.mas.k up
where x (after the eth0) is the number of the alias you wish to use.
You may also need to add an entry to ensure it's made active after a
reboot.
HTH
Best Regards
Daren Russell
End Design Ltd
(remove invalid to reply by email)
------------------------------
From: "Tauno Voipio" <[EMAIL PROTECTED]>
Subject: Re: Win2K messed after RH 6.1 installation
Date: Tue, 03 Apr 2001 15:26:52 GMT
"Christoph Kukulies" <[EMAIL PROTECTED]> wrote in message >
> :> What can I do about it?
>
> : mount the logicals inside. (hda5 to hda8) The mount command was
> : very clear about this.
>
> : If there was a win2k D: it's gone now. You installed linux right over
it.
>
>
> So "over" was correct, above? :-> rats!
> How come? 6.1 disk druid cannot cope with win2k partitions?
>
Next time you're installing the 6.1, use 'text expert' and fdisk to install
the partitions the way you need. Disk Druid does not always behave if there
is already something partitioned / installed in the system.
Tauno Voipio
tauno voipio @ iki fi
------------------------------
From: [EMAIL PROTECTED]
Subject: Re: difference between ext2fs and reiserfs
Date: Tue, 03 Apr 2001 15:33:48 GMT
[EMAIL PROTECTED] (Rod Smith) writes:
> [Posted and mailed]
>
> In article <[EMAIL PROTECTED]>,
> "Andy Walker" <[EMAIL PROTECTED]> writes:
> >
> > While we are on the subject, is there any sort of agreement to a standard
> > new filesystem for Linux? I heard mention of ext3 but how good is it? In my
> > experience, other than the boot limitations which aren't really a problem,
> > ReiserFS is extremely stable and can't be far off from being considered a
> > finished product.
>
> There are at least four journaling filesystems under development for
> Linux:
>
> - ReiserFS -- We've just been discussing this one.
> - ext3fs -- ext2fs with journaling features added.
> - XFS -- SGI's IRIX journaling filesystem, ported to Linux.
> - JFS -- IBM's AIX and OS/2 journaling filesystem, ported to Linux.
>
> Of these, ReiserFS is now included in the kernel (as of kernel 2.4.1).
> I'm not sure what the status is on ext3fs. I hear that XFS is getting
> quite stable, but I've never used it. The last I used JFS, it wasn't
> really a workable Linux filesystem, but that was several months ago; I'm
> sure it's improved since then. I would imagine that the developers of
> all of these filesystems want to see them included in the kernel.
> Unfortunately, both ReiserFS and ext3fs suffer from the same file-
> and partition-size limits as ext2fs. These are starting to become
> issues for some people -- particularly the 4GB file-size
> limit. Therefore, unless those limits are raised, neither of these
> filesystems will really do as more than a stop-gap measure.
Hum? "The same file-size limits as ext2fs" would indicate something
slightly over 4TB.
With 4KB blocks, the limit on filesize is 4TB + 4GB + 4MB + 12 * 4KB.
The _essential_ problem leading to the 2GB (not 4GB) limit you're
thinking of is not with the kernel, but rather with the fact that the
file offset fseek()/ftell() has been treated as a 32 bit signed value.
You might want to look to Alexander Viro's responses to this; should
be readily found by search engines, or at the URL below...
> All of this is IMHO, of course, and is based on my understanding of
> the current situation. If that's wrong (say, if ReiserFS is easily
> extended out to larger partition and file sizes), then my
> conclusions are questionable, as well.
Something like that...
--
(reverse (concatenate 'string "ac.notelrac.teneerf@" "454aa"))
http://vip.hex.net/~cbbrowne/fs.html
Rules of the Evil Overlord #49. "If I learn the whereabouts of the one
artifact which can destroy me, I will not send all my troops out to
seize it. Instead I will send them out to seize something else and
quietly put a Want-Ad in the local paper."
<http://www.eviloverlord.com/>
------------------------------
From: Kent Robotti <[EMAIL PROTECTED]>
Subject: Re: loopback device
Date: Tue, 03 Apr 2001 15:34:42 -0000
Alexander Elsenaar <[EMAIL PROTECTED]> wrote:
>
>
> Has anyone problems with the loopback device with a 2.4.2 kernel?
>
> I recompiled the kernel without any problems and now I am unable to
> mount/access a loopback filesystem. The system completely deadlocks! This
> seemed to be a problem with older kernels but I cannot find any
> registration a this problem with the 2.4.2 kernel.
>
> Anybody?
>
> Alexander Elsenaar
I think the loop problems are fixed with 2.4.3.
------------------------------
From: [EMAIL PROTECTED] (Bill Grzanich)
Subject: Daylight savings time
Date: 3 Apr 2001 15:55:05 GMT
Hello, All.
I have a dual-processor server running RedHat 6.2 (VA Linux), kernel
2.2.14-5.0.14csmp. As delivered, the hardware clock was set to Pacific
Standard Time, so I changed it to Central Standard. Since then, we've had
two instances of time change, fall back and spring forward, and in each
case I've had to manually change the server time, to the amusement of my
WinNT coworkers. The Linux server is used primarily for file and print
sharing, but is also the reference time for the clients... that is, they
execute "net time \\linux02 /set /yes" as part of their logon scripts.
After some research on deja.com, the concensus seems to be that I must set
the hardware clock to UTC (GMT), the timezone to America/Central, and
everything should be fine. So, I tried the following at 4:51pm local time:
hwclock --set --utc --date="4/2/01 21:51:10"
then
hwclock --hctosys
and
timeconfig --utc America/Chicago
which modified the /etc/sysconfig/clock file to:
ZONE="America/Chicago"
UTC=true
ARC=false
When running the date command, however, I get back UTC time with the CDT
timezone!
$ date
Mon Apr 2 21:52:30 CDT 2001
So, I changed things back to the way they were, and now date tells me:
Mon Apr 2 17:04:43 CDT 2001
which is correct. However, I'd like to fix this before we fall back to
CST. What am I missing?
Thanks.
-Bill
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: difference between ext2fs and raiserfs
Date: Tue, 03 Apr 2001 15:59:50 GMT
Rod Smith <[EMAIL PROTECTED]> wrote:
> There are at least four journaling filesystems under development for
> Linux:
> - ReiserFS -- We've just been discussing this one.
> - ext3fs -- ext2fs with journaling features added.
> - XFS -- SGI's IRIX journaling filesystem, ported to Linux.
> - JFS -- IBM's AIX and OS/2 journaling filesystem, ported to Linux.
> Of these, ReiserFS is now included in the kernel (as of kernel 2.4.1).
> I'm not sure what the status is on ext3fs. I hear that XFS is getting
> quite stable, but I've never used it. The last I used JFS, it wasn't
I'm using it experimentally, but over LVM and under RAID so I can't
comment on its stabilty on its own. I've lost the partition a few
times for no discernable reason (unmount, mount, and it refuses
to mount and the recovery tool searches fruitlessly for a superblock).
Probably more to do with lvm than xfs.
I am running xfs 0.9 because that matched the 2.4.0 kernel I had the
best .. and matched the lvm code. The xfs patch is quite invasive,
so it's important to get a good match. The xfs list is very active with
patches going in at about 10 to 20 a day. It'll be stable very soon.
> really a workable Linux filesystem, but that was several months ago; I'm
Ditto.
> sure it's improved since then. I would imagine that the developers of
> all of these filesystems want to see them included in the kernel.
> Unfortunately, both ReiserFS and ext3fs suffer from the same file- and
> partition-size limits as ext2fs. These are starting to become issues
> for some people -- particularly the 4GB file-size limit. Therefore,
It's 2GB. I wasn't aware that the largefile patches to the kernel
didn't affect them (excuse my double negative) nor was I aware of
difficulties with the 3GB limit in 2.4.* in general. I believe that
neither e2fs nor the other fs's have any intrinsic difficulty with
large files (we know e2fs doesn't, because it works on 64bit sparc
with arpbitrary sized files). The obstacle used to be the return
value from the block device seek function in the kernel, which
was 32 bit, and represented the _absolute_ offset from the beginning of
the file. In 2.4.0 that obstacle does not exist. I suspect that there
is no FS imposed limit now ..
For fun, I'll try making a 2.1GB file on xfs under 2.4.0 ...
coming up ...
ditserv2:/mnt/lab% ls -l
total 1028
-rw-r--r-- 1 root 2124414976 Apr 3 17:38 foo
and failed ...
Filesize limit exceeded
Maybe it was my dd that baulked. dd and libc
have to be compiled with largefile support to get the right seek
function. But I KNOW that that's working because I am doing megaseeks
across raw partitions with no trouble in code of mine. Well, the strace
says dd got an error in write ...
read(0, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...,
1048576) = 1048576
write(1, "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0"...,
1048576) = -1 EFBIG (File too large)
--- SIGXFSZ (File size limit exceeded) ---
+++ killed by SIGXFSZ +++
which means that the write to the FS was abended by the FS.
No comment. The same thing occurred on an ext2fs in 2.4.0.
> unless those limits are raised, neither of these filesystems will
> really do as more than a stop-gap measure. OTOH, it's possible to
> convert an existing ext2fs partition into an ext3fs partition without
> losing data, so transitioning to it should be easy. (The others all
I am beginning to feel that this really is the stable route. And if
someone would kindly port e2compr ...
> require backup/make filesystem/restore dances.) Both XFS and JFS are
> 32-bit filesystems with much higher limits on file and partition sizes,
> so they've got potentially bright futures. XFS has somewhat higher
> limits, but they're both so high that I don't see it making much
> difference. Which of these will win out in the end as the "standard"
> Linux filesystem is something I'd not care to predict, although as I
> say, I think XFS and JFS are the strongest long-term contenders.
> All of this is IMHO, of course, and is based on my understanding of the
> current situation. If that's wrong (say, if ReiserFS is easily extended
> out to larger partition and file sizes), then my conclusions are
> questionable, as well.
I wasn't aware that RFS had a 2GB limit, as I mentioned. As we know,
logically e2fs doesn't have one either .. but there appears to be
some magic necessary. I don't know where it has to be applied. Libc?
Tools?
Peter
------------------------------
From: "Peter T. Breuer" <[EMAIL PROTECTED]>
Subject: Re: loopback device
Date: Tue, 03 Apr 2001 15:59:59 GMT
Alexander Elsenaar <[EMAIL PROTECTED]> wrote:
> Has anyone problems with the loopback device with a 2.4.2 kernel?
It doesn't work in any of the 2.4.*. It was supposed to have been fixed
by 2.4.3 but I don't know if it happened. Check.
> I recompiled the kernel without any problems and now I am unable to
> mount/access a loopback filesystem. The system completely deadlocks! This
Right.
> seemed to be a problem with older kernels but I cannot find any
?? It's not a problem of 2.2 kernels.
> registration a this problem with the 2.4.2 kernel.
Mind you, there is always some danger of a vfs lockup against
localhost.
Peter
------------------------------
From: Alexander Elsenaar <[EMAIL PROTECTED]>
Subject: Re: loopback device
Date: Tue, 03 Apr 2001 15:57:13 GMT
Kent Robotti wrote:
> Alexander Elsenaar <[EMAIL PROTECTED]> wrote:
>>
>>
>> Has anyone problems with the loopback device with a 2.4.2 kernel?
>>
>> I recompiled the kernel without any problems and now I am unable to
>> mount/access a loopback filesystem. The system completely deadlocks! This
>> seemed to be a problem with older kernels but I cannot find any
>> registration a this problem with the 2.4.2 kernel.
>>
>> Anybody?
>>
>> Alexander Elsenaar
>
> I think the loop problems are fixed with 2.4.3.
>
Hmmm, misted that one!
------------------------------
** 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
******************************
