ipcop-devel Mailing List for IPCop Firewall

Hi Eric,
Eric Oberlander schrieb:
> This paragraph is from the existing Manual, describing how to Generate
> Root/Host Certificates
> 
> "To create an IPCop's Certificate Authority or CA, enter your CA's
> name in the CA Name box. The name should be different than the IPCop
> machine's host name to avoid confusion. For example, ipcopca for the
> CA and ipcop for the hostname. Then click on the Generate Root/Host
> Certificates  button. "
> 
> Is that correct? You don't seem to need to input an CA name when you
> generate Certificates.
wrong. I'll try:
* To create an IPCop's Root and Host certificates click on the "Generate
Root/Host Certificates" button. This opens a new screen where you need
to enter the details of the certificates. The fields "Organization Name"
and "IPCop's Hostname" are mandatory (where the later is usually already
populated with the hostname or ip of the RED interface), all other
fields are optional. Once you have entered the informations click again
the "Generate Root/Host Certificates" button. *

> But a CA name _is_ required when you Upload a Certificate...
yes, this is for CAs from remote machines, and this name can be
anything, but I would suggest something like this:
if the remote Ipcop is CompanyGateway then name the CA just Company, and
the connection CompanyNet (for a net2net connection).

HTH, Gün.

On Mon, 2009-08-31 at 08:39 +0100, kay wrote:

> -         When I look at ifconfig, I see at the bottom;

> “wan-1.alias: error fetching interface information: Device not
> found”. 

> Any comment?

This is expected behaviour, ifconfig can't correctly handle new-style
aliases created by iproute2. You can use "ip addr list" to view the
aliases with iproute2.

-- 
 Robert Kerr

----- Original Message ----- 
From: "Olaf" <mai...@ba...>
To: "Gilles Espinasse" <g....@fr...>; <ipc...@li...>
Sent: Saturday, August 29, 2009 7:19 PM
Subject: Re: [IPCop-devel] Why /proc/scsi is missing


> Olaf Westrik wrote:
> > John Edwards wrote:
> >
> >> CONFIG_IDE_PROC_FS is set for SPARC, PPC and ALPHA machines, but not
> >> i486. I can not find anywhere in the current SVN code (r3500) where we
> >> use /proc/ide. This may be a case where 1.4.x needed it and the other
> >> architectures have not been updated, or there could be some places in
> >> the installer which really does need it.
> >
> > We do not need it. The installer uses /sys/bus/ide/devices,
> > /sys/block/sd* and /sys/block/sr*.
>
> Is it OK to remove CONFIG_IDE_PROC_FS for ppc and sparc?
>
>
> Olaf

I don't know.
I suppose it should be ok.
I could test, this will require me some time (machines are slow to compile).

There is some other issues I have tried to investigate:
- on sparc it fail to build on final glibc with 1.9.7 toolchain. This had
work until 1.9.5 toolchain.
I suspect it was my upgrade from sarge to lenny that broke the build.
/usr/src/glibc-build/nscd/nscd.o: In function `nscd_open_socket':
/usr/src/glibc-2.7/nscd/nscd.c:419: undefined reference to
`__stack_chk_guard'
/usr/src/glibc-build/nscd/nscd.o: In function `parse_opt':
/usr/src/glibc-2.7/nscd/nscd.c:288: undefined reference to
`__stack_chk_guard'
/usr/src/glibc-build/nscd/connections.o: In function `restart':
/usr/src/glibc-2.7/nscd/connections.c:1193: undefined reference to
`__stack_chk_guard'
/usr/src/glibc-build/nscd/connections.o: In function `nscd_run':
/usr/src/glibc-2.7/nscd/connections.c:1354: undefined reference to
`__stack_chk_guard'
/usr/src/glibc-build/nscd/connections.o: In function `nscd_init':
/usr/src/glibc-2.7/nscd/connections.c:490: undefined reference to
`__stack_chk_guard'
/usr/src/glibc-build/nscd/connections.o:/usr/src/glibc-2.7/nscd/connections.
c:867: more undefined references to `__stack_chk_guard' follow
collect2: ld returned 1 exit status
make[3]: *** [/usr/src/glibc-build/nscd/nscd] Error 1

I try to revert linux-headers to what the value (2.6.27.8) in 1.9.5 with no
change.

If I revert to 1.9.5 toolchain, build is ok.
I am now building with svn reverted to 1.9.5 to verify this is fully the
build machine that change the result.


- on ppc, there is an initrd size limit on network boot (not when booting
from iso)
I have build a kernel with CONFIG_IDE_PROC_FS disabled, I have now to try to
install.

Gilles

On Sun, Aug 30, 2009 at 04:51:03PM +0100, kay wrote:
> I have manage to install.

How?

What was the problem?

How did you fix it?


We still need to know the details of the hardware - CPU, RAM,
disk controllers, RAID, output of 'lspci && lspci -vvn', etc.


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

I have manage to install.


Thanks and regards
 
 
Kay Mbwana
Galileo Angola
 
Tel: 244 222 449105
       244 222 446813
        244 222 446883
 
FAX  244 222 449134
Cell   244 912 500426
 
 
 
-----Original Message-----
From: Eric Oberlander [mailto:eri...@gm...] 
Sent: Sunday, August 30, 2009 3:18 PM
To: ka...@ga...
Cc: John Edwards; ipc...@li...
Subject: Re: [IPCop-devel] FW: Installation Problem -
ipcop-1.9.7-install-cd.i486.iso - No Kernel found.

2009/8/30 kay <ka...@ga...>:
> To John
>
> That error has been solved.
>
> Now I am at a new error as I responded to Eric;
>
>
> Thanks Eric,
>
> I bunt a new CD and managed to move forward a bit.
>
> 1: At load and add a specific module.. I selected skip
> 2: Disk Installation --OK
> 3: Installation of file system .. stops with an error;
>
>                        Tar error
>
> {The actual running command is /bin/tar -c /harddisk -vxpzf
> /cdrom/ipcop-1.9.7.tar.gz}

Try installing again. I had this error on one installation, and tried
agian, and it worked the second time. Sometimes equipment has dust on
it?

Eric

Hi,
Eric Oberlander schrieb:
> Is anybody testing/using VPNs on the new version of IPCop?
yes, tried - but Ipsec is currently broken in 1.9.7 ...

Gün.

2009/8/30 kay <ka...@ga...>:
> To John
>
> That error has been solved.
>
> Now I am at a new error as I responded to Eric;
>
>
> Thanks Eric,
>
> I bunt a new CD and managed to move forward a bit.
>
> 1: At load and add a specific module.. I selected skip
> 2: Disk Installation --OK
> 3: Installation of file system .. stops with an error;
>
>                        Tar error
>
> {The actual running command is /bin/tar -c /harddisk -vxpzf
> /cdrom/ipcop-1.9.7.tar.gz}

Try installing again. I had this error on one installation, and tried
agian, and it worked the second time. Sometimes equipment has dust on
it?

Eric

This paragraph is from the existing Manual, describing how to Generate
Root/Host Certificates

"To create an IPCop's Certificate Authority or CA, enter your CA's
name in the CA Name box. The name should be different than the IPCop
machine's host name to avoid confusion. For example, ipcopca for the
CA and ipcop for the hostname. Then click on the Generate Root/Host
Certificates  button. "

Is that correct? You don't seem to need to input an CA name when you
generate Certificates.

But a CA name _is_ required when you Upload a Certificate...

What's correct?

Eric

On Sun, Aug 30, 2009 at 03:05:02PM +0100, kay wrote:
> To John 
> 
> That error has been solved.
> 
> Now I am at a new error as I responded to Eric;
> 
> 
> Thanks Eric,
> 
> I bunt a new CD and managed to move forward a bit.
> 
> 1: At load and add a specific module.. I selected skip
> 2: Disk Installation --OK
> 3: Installation of file system .. stops with an error;
> 
> 			Tar error
> 
> {The actual running command is /bin/tar -c /harddisk -vxpzf
> /cdrom/ipcop-1.9.7.tar.gz}
> 
> 
> Please help.
<snip> 

Check the MD5SUM values of the downloaded image file and CDROM:

> As Eric Oberlander suggested, to make sure that you have downloaded
> the image file and created the CDROM without error, you should check
> the MD5SUM values of the image file and the CDROM with the values in
> this file:
> 	
> http://downloads.sourceforge.net/project/ipcop/IPCop%20Test%20Versions/IPCop%201.9.7/release-notes-1.9.7.txt


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

To John 

That error has been solved.

Now I am at a new error as I responded to Eric;


Thanks Eric,

I bunt a new CD and managed to move forward a bit.

1: At load and add a specific module.. I selected skip
2: Disk Installation --OK
3: Installation of file system .. stops with an error;

			Tar error

{The actual running command is /bin/tar -c /harddisk -vxpzf
/cdrom/ipcop-1.9.7.tar.gz}


Please help.

Thanks and regards
 
 
Kay Mbwana
Galileo Angola
 
Tel: 244 222 449105
       244 222 446813
        244 222 446883
 
FAX  244 222 449134
Cell   244 912 500426
 
 
 

-----Original Message-----
From: John Edwards [mailto:jo...@co...] 
Sent: Sunday, August 30, 2009 3:01 PM
To: kay
Cc: ipc...@li...
Subject: Re: [IPCop-devel] FW: Installation Problem -
ipcop-1.9.7-install-cd.i486.iso - No Kernel found.

On Sun, Aug 30, 2009 at 01:04:22PM +0100, kay wrote:
> Hi
> 
> Installation of IPCOP 2.0 fails at;
> 
> ISOLINUX 3.8.2 2009-------
> Could not find kernel image: Linux
> Boot:

Is this when you try to boot from the installation CDROM?

As Eric Oberlander suggested, to make sure that you have downloaded
the image file and created the CDROM without error, you should check
the MD5SUM values of the image file and the CDROM with the values in
this file:
	
http://downloads.sourceforge.net/project/ipcop/IPCop%20Test%20Versions/IPCop
%201.9.7/release-notes-1.9.7.txt


If they both match then it is possible that the CDROM is on a
controller that does not have a good driver in IPCop. Do you
know what type of CDROM and what controller it is attached to?


> Regarding the computer,
> 
> Yes we can install Debian Linux without any problem,

If you can still boot usinog Debian or a Live CD (like Knoppix) then
it would be helpful if you could email the results of these commands:
	lspci && lspci -vvn 


> We have SATA disk with SAS 6 adapter . RAID?


Some SAS controllers can do hardware RAID:
	
http://accessories.dell.com/sna/productdetail.aspx?c=ca&cs=cadhs1&l=en&s=dhs
&sku=341-6991&~tab=SpecsTab

Whether you have one that does hardware RAID, and if RAID is
enabled is a question that only you can answer.


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

On Sun, Aug 30, 2009 at 01:04:22PM +0100, kay wrote:
> Hi
> 
> Installation of IPCOP 2.0 fails at;
> 
> ISOLINUX 3.8.2 2009-------
> Could not find kernel image: Linux
> Boot:

Is this when you try to boot from the installation CDROM?

As Eric Oberlander suggested, to make sure that you have downloaded
the image file and created the CDROM without error, you should check
the MD5SUM values of the image file and the CDROM with the values in
this file:
	http://downloads.sourceforge.net/project/ipcop/IPCop%20Test%20Versions/IPCop%201.9.7/release-notes-1.9.7.txt


If they both match then it is possible that the CDROM is on a
controller that does not have a good driver in IPCop. Do you
know what type of CDROM and what controller it is attached to?


> Regarding the computer,
> 
> Yes we can install Debian Linux without any problem,

If you can still boot usinog Debian or a Live CD (like Knoppix) then
it would be helpful if you could email the results of these commands:
	lspci && lspci -vvn 


> We have SATA disk with SAS 6 adapter . RAID?


Some SAS controllers can do hardware RAID:
	http://accessories.dell.com/sna/productdetail.aspx?c=ca&cs=cadhs1&l=en&s=dhs&sku=341-6991&~tab=SpecsTab

Whether you have one that does hardware RAID, and if RAID is
enabled is a question that only you can answer.


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

Thanks Eric,

I bunt a new CD and managed to move forward a bit.

1: At load and add a specific module.. I selected skip
2: Disk Installation --OK
3: Installation of file system .. stops with ann error;

			Tar error

Please assist


Thanks and regards
 
 
Kay Mbwana
Galileo Angola
 
Tel: 244 222 449105
       244 222 446813
        244 222 446883
 
FAX  244 222 449134
Cell   244 912 500426
 
 
 

-----Original Message-----
From: Eric Oberlander [mailto:eri...@gm...] 
Sent: Friday, August 28, 2009 8:05 PM
To: ipc...@li...
Subject: Re: [IPCop-devel] FW: Installation Problem -
ipcop-1.9.7-install-cd.i486.iso - No Kernel found.

2009/8/28 kay <ka...@ga...>:
> Hi,
>
> We have download the ipcop-1.9.7-install-cd.i486.iso
> We have burn a CD.
>
> During installation  we get error - "No kernel found"
>
> We have installed Debian Linux on the same machine but we cannot get
through
> either.
>
>
> We are testing on DELL PowerEdge R300 with two hard disk.
>
> Please advise how we should proceed as we want to participate in testing.
>
> We have other computers that are free.
>
> Thanks and regards
>
>
> Kay Mbwana
> Galileo Angola

Hi

Check the md5sum of the .iso file to make sure it is good, then burn
another copy at slow speed, and try again.

md5sums are listed here:
http://downloads.sourceforge.net/project/ipcop/IPCop%20Test%20Versions/IPCop
%201.9.7/release-notes-1.9.7.txt

Eric

----------------------------------------------------------------------------
--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus
on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
IPCop-devel mailing list
IPC...@li...
https://lists.sourceforge.net/lists/listinfo/ipcop-devel

Hi

Installation of IPCOP 2.0 fails at;



ISOLINUX 3.8.2 2009-------
Could not find kernel image: Linux
Boot:



Regarding the computer,

Yes we can install Debian Linux without any problem,
We have SATA disk with SAS 6 adapter . RAID?

Thanks and regards
 
 
Kay Mbwana
Galileo Angola
 
Tel: 244 222 449105
       244 222 446813
        244 222 446883
 
FAX  244 222 449134
Cell   244 912 500426
 
 
 

-----Original Message-----
From: John Edwards [mailto:jo...@co...] 
Sent: Saturday, August 29, 2009 5:37 PM
To: kay
Cc: ipc...@li...
Subject: Re: [IPCop-devel] FW: Installation Problem -
ipcop-1.9.7-install-cd.i486.iso - No Kernel found.

On Fri, Aug 28, 2009 at 05:22:10PM +0100, kay wrote:
> Hi,
> 
> We have download the ipcop-1.9.7-install-cd.i486.iso
> We have burn a CD.

Thanks for testing the new version.


> During installation  we get error - "No kernel found"

At which stage of the installation do you see this message?


> We have installed Debian Linux on the same machine but we cannot get
through
> either.

Do you mean that you can not install Debian?

If yes, then at which stage of the install does Debian fail and
what is the message?


> We are testing on DELL PowerEdge R300 with two hard disk.

What type of disks - IDE, SATA or SCSI?

What type of hard disk controllers are used (ICH, PERC or SAS)?

Is there any hardware or software RAID configured?


> Please advise how we should proceed as we want to participate in testing.
> 
> We have other computers that are free.

Have you tried the same CD install on another computer?


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

Hi

Is anybody testing/using VPNs on the new version of IPCop?

I'd like to update the Administration Manual sections:
http://www.ipcop.org/2.0.0/en/admin/html/vpns.html with step by step
howtos, or links to existing howtos.

Thanks.

Eric

  Ok Olaf, the recent commit works here. BTW, doing test builds for 
1.4.23cvs reveals the same problem. It's always taken some fiddling to 
get it to report to an external server. This has definitely not a been 
new (2.0) issue from what I can tell. I only had to hit save after 
ensuring the the log to an external server box was ticked and the 
address of my remote syslog server and it started reporting immediately 
on my external server. This is a very nice feature, especially for flash 
installs. So far so good Olaf! Thanks!


-- 
Dave Studeman
http://www.raqcop.com

Olaf Westrik wrote:
> John Edwards wrote:
> 
>> CONFIG_IDE_PROC_FS is set for SPARC, PPC and ALPHA machines, but not
>> i486. I can not find anywhere in the current SVN code (r3500) where we
>> use /proc/ide. This may be a case where 1.4.x needed it and the other
>> architectures have not been updated, or there could be some places in
>> the installer which really does need it.
> 
> We do not need it. The installer uses /sys/bus/ide/devices, 
> /sys/block/sd* and /sys/block/sr*.

Is it OK to remove CONFIG_IDE_PROC_FS for ppc and sparc?


Olaf

Gilles Espinasse wrote:

>> When building the toolchain as non-root, files are installed into
>> tools_$(MACHINE) under this non root owner/group name.
>> But this user name/group is not referenced under chroot
> /etc/{group,passwd}
>> except in one case.
>> User is never know but if build user group is 1000, files look owned by
>> 1000:untrusted
>> ( untrusted:x:1000: is registered in chroot etc/group)
>>
>> Should we not change ownership of every files/directory under
>> tools_$(MACHINE) to root:root before packaging the toolchain (and remove
>> untrusted group from etc/group)?
>> We should care only to avoid following link when changing ownership.

Probably best/easiest to use root:root. We could also add 'build' 
user:group but I do not know if that is worth the trouble?


> Maybe not remove untrusted group 1000 as it look this will add a warning
> when useradd is used without -g (look at lfs page for details)
> 
> Should we not include etc/default/useradd?

useradd is probably one of the least often used programs on an IPCop ;-)
I see no problem to add the file though.


Olaf

On Fri, Aug 28, 2009 at 05:22:10PM +0100, kay wrote:
> Hi,
> 
> We have download the ipcop-1.9.7-install-cd.i486.iso
> We have burn a CD.

Thanks for testing the new version.


> During installation  we get error - "No kernel found"

At which stage of the installation do you see this message?


> We have installed Debian Linux on the same machine but we cannot get through
> either.

Do you mean that you can not install Debian?

If yes, then at which stage of the install does Debian fail and
what is the message?


> We are testing on DELL PowerEdge R300 with two hard disk.

What type of disks - IDE, SATA or SCSI?

What type of hard disk controllers are used (ICH, PERC or SAS)?

Is there any hardware or software RAID configured?


> Please advise how we should proceed as we want to participate in testing.
> 
> We have other computers that are free.

Have you tried the same CD install on another computer?


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#

----- Original Message ----- 
From: "Gilles Espinasse" <g....@fr...>
To: "IPCOP devel" <ipc...@li...>
Sent: Saturday, August 29, 2009 8:14 AM
Subject: [2.0] build user is not a valid user inside the chroot


> When building the toolchain as non-root, files are installed into
> tools_$(MACHINE) under this non root owner/group name.
> But this user name/group is not referenced under chroot
/etc/{group,passwd}
> except in one case.
> User is never know but if build user group is 1000, files look owned by
> 1000:untrusted
> ( untrusted:x:1000: is registered in chroot etc/group)
>
> Should we not change ownership of every files/directory under
> tools_$(MACHINE) to root:root before packaging the toolchain (and remove
> untrusted group from etc/group)?
> We should care only to avoid following link when changing ownership.
>
> Gilles
>
Maybe not remove untrusted group 1000 as it look this will add a warning
when useradd is used without -g (look at lfs page for details)

Should we not include etc/default/useradd?
We care to change this file (switching off mailbox creation) in lfs/shadow
but did not include the resulting file.

Gilles

Olaf Westrik schrieb:
> v1.9.7 is *TESTING* material only, *DO NOT* use in a productive environment.
a follow-up to my other test results: just my Alix box did not respond
now for the third time, so I thought I mention it here; unfortunately I
can say nothing more about it since also serial console did not longer
respond, and I had to reset the box; all logs seem gone since flash
install ...
This is standard 1.9.7 flash install via PXE, no addons installed (and
no wireless card!), except for the LED control kernel modules (but 2 of
these total freezes already happened before I added the LED stuff).

Günter.

When building the toolchain as non-root, files are installed into
tools_$(MACHINE) under this non root owner/group name.
But this user name/group is not referenced under chroot /etc/{group,passwd}
except in one case.
User is never know but if build user group is 1000, files look owned by
1000:untrusted
( untrusted:x:1000: is registered in chroot etc/group)

Should we not change ownership of every files/directory under
tools_$(MACHINE) to root:root before packaging the toolchain (and remove
untrusted group from etc/group)?
We should care only to avoid following link when changing ownership.

Gilles

2009/8/28 kay <ka...@ga...>:
> Hi,
>
> We have download the ipcop-1.9.7-install-cd.i486.iso
> We have burn a CD.
>
> During installation  we get error - "No kernel found"
>
> We have installed Debian Linux on the same machine but we cannot get through
> either.
>
>
> We are testing on DELL PowerEdge R300 with two hard disk.
>
> Please advise how we should proceed as we want to participate in testing.
>
> We have other computers that are free.
>
> Thanks and regards
>
>
> Kay Mbwana
> Galileo Angola

Hi

Check the md5sum of the .iso file to make sure it is good, then burn
another copy at slow speed, and try again.

md5sums are listed here:
http://downloads.sourceforge.net/project/ipcop/IPCop%20Test%20Versions/IPCop%201.9.7/release-notes-1.9.7.txt

Eric

Hi,

We have download the ipcop-1.9.7-install-cd.i486.iso
We have burn a CD.

During installation  we get error - "No kernel found"

We have installed Debian Linux on the same machine but we cannot get through
either.


We are testing on DELL PowerEdge R300 with two hard disk.

Please advise how we should proceed as we want to participate in testing.

We have other computers that are free.

Thanks and regards
 
 
Kay Mbwana
Galileo Angola

Olaf Westrik wrote:

> David W Studeman wrote:
> 
>> Here is what I see in the logs on the first try after a fresh install
>> bootup after hitting save in the log settings external syslog server
>> section in the web gui:
>> 
>> Aug 27 00:39:54 testcop syslogd 1.5.0: restart.
>> Aug 27 00:39:54 testcop syslogd: sendto: Bad file descriptor
> 
> I think the problem here is that we run syslogd as user syslogd (not as
> root).
> restartsyslogd restarts syslogd by sending HUP, syslogd then tries to
> open a UDP port which will fail (ports < 1024 require root user).
> 
> 
> To work around this there are several solutions:
>   1) remove the remote syslog feature
>   2) run syslogd as root
>   3) restartsyslogd kills and starts syslogd instead of sending HUP
>   4) as 3 but only kill and start if remote syslog is enabled

 Number 4 sounds the best to me at least.  
> 
>> And then after rebooting it appears to try to start syslog to the
>> external server long before networking is started:
>> 
>> Aug 27 00:45:25 testcop syslogd 1.5.0: restart.
>> Aug 27 00:45:25 testcop syslogd: sendto: Network is unreachable
>> Aug 27 00:45:25 testcop kernel: klogd 1.5.0, log source = /proc/kmsg
>> started.
>> 
>> Then after it is up and running it will work by hitting save in the log
>> settings external syslog server section. Notice that I did it about a
>> minute later extra entries are included to show that there were no
>> further errors at this point:
>> 
>> Aug 27 00:45:46 testcop ntpd[819]: kernel time sync status change 0001
>> Aug 27 00:46:19 testcop syslogd 1.5.0: restart.
>> Aug 27 00:48:34 testcop kernel: GREEN REJECT IN=lan-1 OUT=
>> MAC=ff:ff:ff:ff:ff:ff:00:30:1b:37:80:f1:08:00
>> -
> 
> Interesting. I'll need to think about how to best solve this a bit more.
> I do not think changing the order in rc.sysinit is an option as that
> would keep logmessages from network initialization out of
> /var/log/messages.

Yup, changing the order is probably a very bad idea. Good system logs are 
invaluable from as early on in the boot process as is possible.

-- 
Dave
http://www.raqcop.com

On Thu, Aug 27, 2009 at 07:24:33PM +0200, Olaf Westrik wrote:
> John Edwards wrote:
<snip>
>> An alternative fix would be to get a list of SCSI devices or drives
>> out of the /sys filesystem. The /sys filesystem is a more complex,
>> but a starting place could to read:
>> 	/sys/class/scsi_disk/*/device/model
>>
>> You would have to process the path and contents to get the device
>> names and info about the SCSI channel, ID, LUN, etc.
>>
>> I think /proc/scsi/scsi is easier, as long as the 2.6 kernel continue
>> to support it.
>
> How about using sdparm?
> For SATA and IDE we use hdparm in sysinfo.cgi,

hdparm fails for SCSI devices with:
	HDIO_DRIVE_CMD(identify) failed: Invalid argument
or on a virtual machine:
	HDIO_DRIVE_CMD(identify) failed: Invalid Exchange
Which is as documented, because it is an ATA operation.

sdparm could be made to work if it was compiled and installed,
and a new option added to /usr/local/bin/sysinfo for SCSI disks.


> much more readable than  
> anything /sys or /proc can deliver without manual searching, tweaking,  
> combining and interpreting.

/proc/scsi/scsi has a very simple layout:
--------
$ more /proc/scsi/scsi 
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
  Vendor: ATA      Model: SAMSUNG HD103SI  Rev: 1AG0
  Type:   Direct-Access                    ANSI  SCSI revision: 05
Host: scsi1 Channel: 00 Id: 00 Lun: 00
  Vendor: ATA      Model: SAMSUNG HD103SI  Rev: 1AG0
  Type:   Direct-Access                    ANSI  SCSI revision: 05
Host: scsi10 Channel: 00 Id: 00 Lun: 00
  Vendor: SAMSUNG  Model: HD103UJ          Rev:     
  Type:   Direct-Access                    ANSI  SCSI revision: 02
--------


On a SATA drive accessed as a SCSI device, sdparm reports:
--------
$ sudo sdparm -i /dev/sda
[sudo] password for john: 
    /dev/sda: ATA       SAMSUNG HD103SI   1AG0
Device identification VPD page:
  Addressed logical unit:
    designator type: vendor specific [0x0],  code set: ASCII
 00     53 31 56 53 4a 39 30 53  35 36 32 32 37 37 20 20    S1VSJ90S562277  
 10     20 20 20 20                                                         
    designator type: T10 vendor identification,  code set: ASCII
      vendor id: ATA     
      vendor specific: SAMSUNG HD103SI                         S1VSJ90S562277
--------

Unfortunately I don't have access to a test machine with a real SCSI
disk, sdparm and a recent 2.6.x kernel, so this may be different with
a real SCSI device.


Which is a lot less info than hdparm on the same drive:
--------
$ sudo hdparm -i /dev/sda

/dev/sda:

 Model=SAMSUNG HD103SI                         , FwRev=1AG01118, SerialNo=S1VSJ90S562277      
 Config={ Fixed }
 RawCHS=16383/16/63, TrkSize=34902, SectSize=554, ECCbytes=4
 BuffType=DualPortCache, BuffSize=32767kB, MaxMultSect=16, MultSect=?0?
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=1953525168
 IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4 
 DMA modes:  mdma0 mdma1 mdma2 
 UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6 
 AdvancedPM=yes: disabled (255) WriteCache=enabled
 Drive conforms to: unknown:  ATA/ATAPI-3,4,5,6,7

 * signifies the current active mode
--------


-- 
#---------------------------------------------------------#
|    John Edwards   Email: jo...@co...    |
#---------------------------------------------------------#