devil-linux-develop Mailing List for Devil-Linux (Page 42)

Quoting Bruce Smith <bw...@ar...>:

> I noticed that it looks for configs on /dev/hd* drives before /dev/sd*
> drives, which means it checks IDE hard drives before USB flash drives.
>
> Is that the way it should work Heiko?
>
> IIRC, it also checks for the config on the CD before external drives
> too, which means you can't override a config burned on the CD.

I'm open on what we check first, you can change it if you want.

-- 

Regards
   Heiko Zuerker
   http://www.devil-linux.org

I noticed that it looks for configs on /dev/hd* drives before /dev/sd*
drives, which means it checks IDE hard drives before USB flash drives.

Is that the way it should work Heiko?

IIRC, it also checks for the config on the CD before external drives
too, which means you can't override a config burned on the CD.

 - BS

I still have not found a floppy disk, but I tried another computer;
everything works as expected, that is the system booting properly
after lots of grumbling on screen because I don't have a configuration
media.

-- 
Arnaud

On Mon, November 5, 2007 15:33, Arnaud Gomes-do-Vale wrote:
> Bruce Smith <bw...@ar...> writes:
>
>
>> The thing I've found with floppies is they are very unreliable.  I
>> actually had to find a different machine because two of the three
>> machines I tried, the floppy drive didn't work.  And there is always
>> the high rate of media failure problems.  Perhaps the original poster is
>>  having a related problem?
>
> I didn't have a floppy in the drive, and AFAIR the drive light stayed
> off. I will try to find out more tomorrow; at the moment, as Devil issues
> proper seem to have been ruled out, I suspect either broken or badly
> supported hardware. More probably broken as the drive has probably not
> been used in years.

Yeah my floppy drives get rarely used either.
The only 'floppy drives' I'm using are the virtual ones in VMWare. ;-)

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Bruce Smith <bw...@ar...> writes:

> The thing I've found with floppies is they are very unreliable.  I
> actually had to find a different machine because two of the three
> machines I tried, the floppy drive didn't work.  And there is always the
> high rate of media failure problems.  Perhaps the original poster is
> having a related problem?

I didn't have a floppy in the drive, and AFAIR the drive light stayed
off. I will try to find out more tomorrow; at the moment, as Devil
issues proper seem to have been ruled out, I suspect either broken or
badly supported hardware. More probably broken as the drive has
probably not been used in years.

-- 
Arnaud

> >>>>> The first problem I ran into: when I boot the test release on a
> >>>>> PC
> >>>>> without etc.tar.bz2 (with an existing CentOS installation on its
> >>>>> hard disk, but this should not come into play), the boot hangs
> >>>>> seemingly forever when trying to load the config from /dev/fd0 (I
> >>>>> waited for 10-15 minutes before
> >>>>> rebooting). It boots correctly if I insert a USB pen drive with an
> >>>>>  etc.tar.bz2 file from a box running Devil 1.2.x.
> >>>>
> >>>> Did you have a floppy in the drive? If yes, was it formated?
> >>>
> >>> No. I am not even sure the floppy drive still works, all I know is
> >>> that it is in the box.
> >>
> >> OK.
> >> Did anybody test the 1.3 boot with a floppy disk on real hardware (it
> >> works on my vmware server)?
> >
> > I used the same ISO that I uploaded to the DL FTP site on two real PC's.
> > One was a 1Ghz box, the other is a 933Mhz box.  Both have floppy drives
> > and IDE hard drives.  They both also have a floppy drive, although I'm
> > using a USB flash drive to save my configs.  I didn't have any problems
> > creating a new config on my USB, and it loads fine after it was created.
> >
> > I actually put the 933Mhz box running DL 1.3 into production last Friday
> > running a firewall and LAMP server at work.
> >
> > For a limited time, I have still the 1Ghz box available for testing, if
> > you have anything specific you'd like me to try.
> 
> Try using the floppy for the config media, just to confirm that it works.

I tried it, and it works fine for me.

I started out with a blank floppy (no USB drive), and it found and wrote
the new config to floppy.  Then I rebooted, and it found and loaded the
config without a problem.

The thing I've found with floppies is they are very unreliable.  I
actually had to find a different machine because two of the three
machines I tried, the floppy drive didn't work.  And there is always the
high rate of media failure problems.  Perhaps the original poster is
having a related problem?

 - BS

On Mon, November 5, 2007 10:38, Bruce Smith wrote:
>>>>> The first problem I ran into: when I boot the test release on a
>>>>> PC
>>>>> without etc.tar.bz2 (with an existing CentOS installation on its
>>>>> hard disk, but this should not come into play), the boot hangs
>>>>> seemingly forever when trying to load the config from /dev/fd0 (I
>>>>> waited for 10-15 minutes before
>>>>> rebooting). It boots correctly if I insert a USB pen drive with an
>>>>>  etc.tar.bz2 file from a box running Devil 1.2.x.
>>>>
>>>> Did you have a floppy in the drive? If yes, was it formated?
>>>>
>>>>
>>>
>>> No. I am not even sure the floppy drive still works, all I know is
>>> that it is in the box.
>>
>> OK.
>> Did anybody test the 1.3 boot with a floppy disk on real hardware (it
>> works on my vmware server)?
>
> I used the same ISO that I uploaded to the DL FTP site on two real PC's.
> One was a 1Ghz box, the other is a 933Mhz box.  Both have floppy drives
> and IDE hard drives.  They both also have a floppy drive, although I'm
> using a USB flash drive to save my configs.  I didn't have any problems
> creating a new config on my USB, and it loads fine after it was created.
>
> I actually put the 933Mhz box running DL 1.3 into production last Friday
> running a firewall and LAMP server at work.
>
> For a limited time, I have still the 1Ghz box available for testing, if
> you have anything specific you'd like me to try.

Try using the floppy for the config media, just to confirm that it works.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

> >>> The first problem I ran into: when I boot the test release on a PC
> >>> without etc.tar.bz2 (with an existing CentOS installation on its hard
> >>> disk, but this should not come into play), the boot hangs seemingly
> >>> forever when trying to load the config from /dev/fd0 (I waited for
> >>> 10-15 minutes before
> >>> rebooting). It boots correctly if I insert a USB pen drive with an
> >>> etc.tar.bz2 file from a box running Devil 1.2.x.
> >>
> >> Did you have a floppy in the drive? If yes, was it formated?
> >>
> >
> > No. I am not even sure the floppy drive still works, all I know is
> > that it is in the box.
> 
> OK.
> Did anybody test the 1.3 boot with a floppy disk on real hardware (it
> works on my vmware server)?

I used the same ISO that I uploaded to the DL FTP site on two real PC's.
One was a 1Ghz box, the other is a 933Mhz box.  Both have floppy drives
and IDE hard drives.  They both also have a floppy drive, although I'm
using a USB flash drive to save my configs.  I didn't have any problems
creating a new config on my USB, and it loads fine after it was created.

I actually put the 933Mhz box running DL 1.3 into production last Friday
running a firewall and LAMP server at work.

For a limited time, I have still the 1Ghz box available for testing, if
you have anything specific you'd like me to try.

 - BS

On Mon, November 5, 2007 09:51, Arnaud Gomes-do-Vale wrote:
> "Heiko Zuerker" <he...@zu...> writes:
>
>
>>> The first problem I ran into: when I boot the test release on a PC
>>> without etc.tar.bz2 (with an existing CentOS installation on its hard
>>> disk, but this should not come into play), the boot hangs seemingly
>>> forever when trying to load the config from /dev/fd0 (I waited for
>>> 10-15 minutes before
>>> rebooting). It boots correctly if I insert a USB pen drive with an
>>> etc.tar.bz2 file from a box running Devil 1.2.x.
>>
>> Did you have a floppy in the drive? If yes, was it formated?
>>
>
> No. I am not even sure the floppy drive still works, all I know is
> that it is in the box.

OK.
Did anybody test the 1.3 boot with a floppy disk on real hardware (it
works on my vmware server)?

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

"Heiko Zuerker" <he...@zu...> writes:

>> The first problem I ran into: when I boot the test release on a PC
>> without etc.tar.bz2 (with an existing CentOS installation on its hard disk,
>> but this should not come into play), the boot hangs seemingly forever when
>> trying to load the config from /dev/fd0 (I waited for 10-15 minutes before
>> rebooting). It boots correctly if I insert a USB pen drive with an
>> etc.tar.bz2 file from a box running Devil 1.2.x.
>
> Did you have a floppy in the drive? If yes, was it formated?

No. I am not even sure the floppy drive still works, all I know is
that it is in the box.

-- 
Arnaud

On Mon, November 5, 2007 05:49, Arnaud Gomes-do-Vale wrote:
> Hi,
>
>
> Bruce Smith <bw...@ar...> writes:
>
>
>> I compiled the latest 1.3-beta last night with the i686 server options
>> and uploaded it to the DL FTP server for people to test:
>
> The first problem I ran into: when I boot the test release on a PC
> without etc.tar.bz2 (with an existing CentOS installation on its hard disk,
> but this should not come into play), the boot hangs seemingly forever when
> trying to load the config from /dev/fd0 (I waited for 10-15 minutes before
> rebooting). It boots correctly if I insert a USB pen drive with an
> etc.tar.bz2 file from a box running Devil 1.2.x.
>
> I have not been able to test it further yet.

Did you have a floppy in the drive? If yes, was it formated?

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi,

Bruce Smith <bw...@ar...> writes:

> I compiled the latest 1.3-beta last night with the i686 server options
> and uploaded it to the DL FTP server for people to test:

The first problem I ran into: when I boot the test release on a PC
without etc.tar.bz2 (with an existing CentOS installation on its hard
disk, but this should not come into play), the boot hangs seemingly
forever when trying to load the config from /dev/fd0 (I waited for
10-15 minutes before rebooting). It boots correctly if I insert a USB
pen drive with an etc.tar.bz2 file from a box running Devil 1.2.x.

I have not been able to test it further yet.

-- 
Arnaud

I compiled the latest 1.3-beta last night with the i686 server options
and uploaded it to the DL FTP server for people to test:

ftp://ftp.devil-linux.org/pub/devel/testing/

 - BS

On Thu, November 1, 2007 15:16, Serge Leschinsky wrote:
> Bruce Smith wrote:
>
>>>> Is it the /probe_all_scsi_modules script that crashes your system?
>>>>
>>>>
>>> Yes. To be more exact, kernel crashes during mount attempt after
>>> loading all sata/pata/scsi modules. If I load only one module
>>> (ata_piix in my case)
>>> everything is ok.
>>
>> It's probably not a good idea to just load all the modules and see
>> which ones stick.  It'd be better to examine the hardware and only load
>> the correct modules -- although I'm not sure how to code that.
>
> Agree. Heiko said that it may be done by mdev using. I'd like to play
> with it and implement correct hardware detection.

mdev is already in the current initrd for 1.3. It's the program which
creates all the devices in /dev/ for us.
So if it doesn't load the module already, you're screwed. ;-)

The problem with the hardware detection is pretty big. I know there are a
few projects out there, but if they're not actively maintained the won't
help us much.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Bruce Smith wrote:
>>> Is it the /probe_all_scsi_modules script that crashes your system?
>>>
>> Yes. To be more exact, kernel crashes during mount attempt after loading all
>> sata/pata/scsi modules. If I load only one module (ata_piix in my case)
>> everything is ok.
> 
> It's probably not a good idea to just load all the modules and see which
> ones stick.  It'd be better to examine the hardware and only load the
> correct modules -- although I'm not sure how to code that.

Agree. Heiko said that it may be done by mdev using. I'd like to play with it
and implement correct hardware detection.

--
Sincerely,
Serge Leschinsky

> > Is it the /probe_all_scsi_modules script that crashes your system?
> > 
> Yes. To be more exact, kernel crashes during mount attempt after loading all
> sata/pata/scsi modules. If I load only one module (ata_piix in my case)
> everything is ok.

It's probably not a good idea to just load all the modules and see which
ones stick.  It'd be better to examine the hardware and only load the
correct modules -- although I'm not sure how to code that.

 - BS

Bruce,

> 
> Is it the /probe_all_scsi_modules script that crashes your system?
> 
Yes. To be more exact, kernel crashes during mount attempt after loading all
sata/pata/scsi modules. If I load only one module (ata_piix in my case)
everything is ok.

-- 
Sincerely,
Serge Leschinsky

> >>> I just stuck an IDE hard drive in the box, and it shows up as hda:
> >> It probably depends on the chipset.
> >> I guess the one Serge is using is only supported by a newer driver which
> >> is only in libata.
> > 
> > My emulated IDE CDROM drive in VMware-workstation also shows up
> > as /dev/hda (FWIW).
> 
> Yes, you are absolutely right. Your hardware doesn't require libata, so you see
> devices in "non scsi" notation. As for me, my box requires ata_piix module not
> only for sata but for pata devices as well. Till I don't see your messages I
> thought that _all_ pata devices are working through libata in 2.6.23.1, that's
> why I was so surprised when you post your configuration. Sorry for confusing.

OK, makes sense.

> I found the reason of crashes - it's my own addition. I added it to be able to
> load my box without usb-drive. It may be useful but I haven't checked in the
> following code, because the executing of it crashes kernel every time.
> 
> PS. construction [ -n "$CONFIG_SRC_LIST"  ] doesn't work for some reason.
> ----------------------------------------------------------------------------------
>  diff -aur ./linuxrc.orig ./linuxrc
> --- ./linuxrc.orig      2007-10-25 15:37:57.000000000 +0400
> +++ ./linuxrc   2007-11-01 12:49:21.000000000 +0300
> @@ -231,6 +223,17 @@
> 
>  CDROMS=$(grep -i "drive name" /proc/sys/dev/cdrom/info)
> 
> +if  echo $CONFIG_SRC_LIST | grep "/dev" > /dev/null ; then
> +    echo "Devices: $CONFIG_SRC_LIST"
> +else
> +    echo "No devices were found"
> +    echo -n "Would you like to probe for SCSI/ATA controllers now?"
> +    if ask_yes_no ; then
> +    /probe_all_scsi_modules
> +    fi
> +fi

Is it the /probe_all_scsi_modules script that crashes your system?

 - BS

Bruce,

Bruce Smith wrote:
>>> I just stuck an IDE hard drive in the box, and it shows up as hda:
>> It probably depends on the chipset.
>> I guess the one Serge is using is only supported by a newer driver which
>> is only in libata.
> 
> My emulated IDE CDROM drive in VMware-workstation also shows up
> as /dev/hda (FWIW).

Yes, you are absolutely right. Your hardware doesn't require libata, so you see
devices in "non scsi" notation. As for me, my box requires ata_piix module not
only for sata but for pata devices as well. Till I don't see your messages I
thought that _all_ pata devices are working through libata in 2.6.23.1, that's
why I was so surprised when you post your configuration. Sorry for confusing.

I found the reason of crashes - it's my own addition. I added it to be able to
load my box without usb-drive. It may be useful but I haven't checked in the
following code, because the executing of it crashes kernel every time.

PS. construction [ -n "$CONFIG_SRC_LIST"  ] doesn't work for some reason.
----------------------------------------------------------------------------------
 diff -aur ./linuxrc.orig ./linuxrc
--- ./linuxrc.orig      2007-10-25 15:37:57.000000000 +0400
+++ ./linuxrc   2007-11-01 12:49:21.000000000 +0300
@@ -231,6 +223,17 @@

 CDROMS=$(grep -i "drive name" /proc/sys/dev/cdrom/info)

+if  echo $CONFIG_SRC_LIST | grep "/dev" > /dev/null ; then
+    echo "Devices: $CONFIG_SRC_LIST"
+else
+    echo "No devices were found"
+    echo -n "Would you like to probe for SCSI/ATA controllers now?"
+    if ask_yes_no ; then
+    /probe_all_scsi_modules
+    fi
+fi
+
+
 # search for configuration information
 DL_CONFIG_SOURCE=""
 DL_CONFIG_FILE=""
-----------------------------------------------------------------------------------

-- 
Sincerely
Serge Leschinsky

Hey,

On Wed, October 31, 2007 06:38, Arnaud Gomes-do-Vale wrote:
> I am currently building 1.3. I had to install python manually in the
> chroot for building jailkit. Is lfssystem-SVN-20070314-cleaned.tar.bz2 the
> right archive?

Yes that's the right one.

Simply selecting Python in "make menuconfig" will resolve this issue. Or
just edit the .config file directly and make sure you got CONFIG_PYTHON=y
.


-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

Hi,

I am currently building 1.3. I had to install python manually in the
chroot for building jailkit. Is lfssystem-SVN-20070314-cleaned.tar.bz2
the right archive?

-- 
Arnaud

> > I just stuck an IDE hard drive in the box, and it shows up as hda:
>
> It probably depends on the chipset.
> I guess the one Serge is using is only supported by a newer driver which
> is only in libata.

My emulated IDE CDROM drive in VMware-workstation also shows up
as /dev/hda (FWIW).

 - BS

On Tue, October 30, 2007 14:15, Bruce Smith wrote:
>>>> OK, I have the latest 1.3 running on a real PC, connected to the
>>>> Internet with a static IP.  It didn't have any problem finding my
>>>> IDE
>>>> CDROM drive (/dev/hdc).
>>>>
>>> /dev/hdc ? Are you sure ? I'm requesting because PATA devices which
>>> work through libata represent themselvs as SCSI (SATA) /dev/sdx...
>>
>> It says hdc when it boots, and so does 'df' (copy/paste):
>>
>>
>> root@Devil:~ # df
>> Filesystem           1K-blocks      Used Available Use% Mounted on
>> rootfs                  266046    266046         0 100% / none
>> 257504      9132    248372   4% /shm
>> /dev/hdc                266046    266046         0 100% /
>> none                    257504      9132    248372   4% /dev root@Devil:~
>> # cat /DEVIL-LINUX
>> 1.3.4-2007-10-30
>> root@Devil:~ # uname -a
>> Linux Devil 2.6.23.1 #1 SMP Tue Oct 30 08:32:52...
>>
>
>
> I just stuck an IDE hard drive in the box, and it shows up as hda:
>
>
>
> root@Devil:~ # fdisk -l /dev/hda
>
>
> Disk /dev/hda: 4311 MB, 4311982080 bytes
> 255 heads, 63 sectors/track, 524 cylinders
> Units = cylinders of 16065 * 512 = 8225280 bytes
>
>
> Device Boot      Start         End      Blocks   Id  System
> /dev/hda1               1         524     4208998+  8e  Linux LVM

It probably depends on the chipset.
I guess the one Serge is using is only supported by a newer driver which
is only in libata.

-- 

Regards
  Heiko Zuerker
  http://www.devil-linux.org

> > > OK, I have the latest 1.3 running on a real PC, connected to the
> > > Internet with a static IP.  It didn't have any problem finding my IDE
> > > CDROM drive (/dev/hdc).
> > /dev/hdc ? Are you sure ? I'm requesting because PATA devices which work through
> >  libata represent themselvs as SCSI (SATA) /dev/sdx...
> 
> It says hdc when it boots, and so does 'df' (copy/paste):
> 
> root@Devil:~ # df
> Filesystem           1K-blocks      Used Available Use% Mounted on
> rootfs                  266046    266046         0 100% /
> none                    257504      9132    248372   4% /shm
> /dev/hdc                266046    266046         0 100% /
> none                    257504      9132    248372   4% /dev
> root@Devil:~ # cat /DEVIL-LINUX  
> 1.3.4-2007-10-30
> root@Devil:~ # uname -a
> Linux Devil 2.6.23.1 #1 SMP Tue Oct 30 08:32:52...


I just stuck an IDE hard drive in the box, and it shows up as hda:


root@Devil:~ # fdisk -l /dev/hda

Disk /dev/hda: 4311 MB, 4311982080 bytes
255 heads, 63 sectors/track, 524 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

   Device Boot      Start         End      Blocks   Id  System
/dev/hda1               1         524     4208998+  8e  Linux LVM


 - BS

> yes I have the same version.
> I checked my config file and I already had the entries commented out,
> that's probably why it worked.
> 
> Maybe we should comment it out in our default config (do you have time to
> do it?).

They are commented out by default, and they should be.  I had to
UNCOMMENT them (one or the other, but not both) to make sshd start.

Also, when I uncommented 'AddressFamily', I had to change the parameter
to 'inet' (IPV4), from the default setting of 'any'.

This is some weird IPv4 vs. IPv6 problem with SSH and the way we have
our network configured.

 - BS


> >>> I have it setup with the standard 2NIC firewall, running a DHCP
> >>> server on the internal network.  NAT works fine, as I can connect to
> >>> the internet with my laptop, using the 1.3 DL box as my firewall.
> >>>
> >>> I am having a weird problem with networking.  sshd wouldn't start.
> >>> It
> >>> kept complaining that the port 0.0.0.0:22 was already in use.  I
> >>> changed the port it listens on to a high number, and it still gave the
> >>> same error (with the new port number).  'netstat -aepn' shows NO TCP
> >>> connections listening at all, only UDP and RAW listening for dhcpd.
> >>>
> >>> I changed sshd_config to manually specify the IP addresses of both
> >>> NIC's
> >>> and now sshd starts fine.
> >>>
> >>> I removed IPV6 routing from the config, with no difference.
> >>>
> >>>
> >>> Something is really weird ...  Ideas???
> >>>
> >>
> >> I'm using samba and ssh on my virtual machine which is running DL 1.3,
> >> no such trouble. Although I don't have the firewall enabled.
> >>
> >> Can you disconnect the box from the web and try without firewall?
> >>
> >
> > I just tried starting sshd in my VMware copy (brand new default config),
> > and got the same error in tty10:
> >
> > Bind to port 22 on 0.0.0.0 failed: Address already in use.
> >
> >
> > I did learn that if I uncomment this line in sshd_config, it works:
> >
> >
> > ListenAddress ::
> >
> >
> > OR, if I add this line in sshd_config (instead), it works:
> >
> >
> > AddressFamily inet
> >
> >
> > OpenSSH problem?
> > Are you running the same version of ssh in your DL 1.3 (4.7p1)?
> >
> >
> > - BS