headers
David Brownlee | 2 Mar 2009 10:22
Picon

Re: Netbooting a machine w/ a disk

On Sat, 28 Feb 2009, Dustin Marquess wrote:

> Awesome.  Out of shear luck I did just that last night and started the
> kernel compile before I went to bed.  Thanks for the confirmation :).
>
> If I want to use RAIDframe in a RAID-1 setup, do I change the root
> setting from sd0 to raid0?

 	If you set the raid to autoconfigure as a root filesystem
 	(raidctl raid0 -A root) then a kernel will automatically
 	use it as the root filesystem, at least for a standard
 	kernel without a hardcoded root device. I've not tried it
 	with a kernel which had hardcoded the root device - let us
 	know what happens :)

--

-- 
 		David/absolute       -- www.NetBSD.org: No hype required --
Permalink | Reply | 5 comments |
headers
Dustin Marquess | 2 Mar 2009 21:42
Picon

Re: Netbooting a machine w/ a disk

On Mon, Mar 2, 2009 at 3:22 AM, David Brownlee <abs <at> netbsd.org> wrote:
> On Sat, 28 Feb 2009, Dustin Marquess wrote:
>
>> Awesome.  Out of shear luck I did just that last night and started the
>> kernel compile before I went to bed.  Thanks for the confirmation :).
>>
>> If I want to use RAIDframe in a RAID-1 setup, do I change the root
>> setting from sd0 to raid0?
>
>        If you set the raid to autoconfigure as a root filesystem
>        (raidctl raid0 -A root) then a kernel will automatically
>        use it as the root filesystem, at least for a standard
>        kernel without a hardcoded root device. I've not tried it
>        with a kernel which had hardcoded the root device - let us
>        know what happens :)

Honestly I went ahead and rebuilt a kernel hardcoded to raid0a to be
on the safe side, and it worked great.  I still have the old one
backed up however that I could try if you really want me to, but
honestly I'm slightly scared :).

Only weird thing that I've noticed is that once I'm in NetBSD, if I
reboot/halt back to SRM and try to boot again, BOOTP just sits there
and times out, almost like network connectivity isn't working.  But if
I 'init', it works again.  Very strange..

Thanks for all of the help!

-Dustin
(Continue reading)

Permalink | Reply | 4 comments |
headers
David Brownlee | 2 Mar 2009 23:58
Picon

Re: Netbooting a machine w/ a disk

On Mon, 2 Mar 2009, Dustin Marquess wrote:

>>        If you set the raid to autoconfigure as a root filesystem
>>        (raidctl raid0 -A root) then a kernel will automatically
>>        use it as the root filesystem, at least for a standard
>>        kernel without a hardcoded root device. I've not tried it
>>        with a kernel which had hardcoded the root device - let us
>>        know what happens :)
>
> Honestly I went ahead and rebuilt a kernel hardcoded to raid0a to be
> on the safe side, and it worked great.  I still have the old one
> backed up however that I could try if you really want me to, but
> honestly I'm slightly scared :).

 	Heh, give is a span next time you need to reboot for any
 	other reason maybe.

> Only weird thing that I've noticed is that once I'm in NetBSD, if I
> reboot/halt back to SRM and try to boot again, BOOTP just sits there
> and times out, almost like network connectivity isn't working.  But if
> I 'init', it works again.  Very strange..

 	It sounds like NetBSD is leaving the network interface in a
 	state that SRM doesn't quite handle. Out of curiosity what
 	happens if you delete the network interface and shutdown
 	the interface before halting - 'ifconfig tlp0 delete'
 	'ifconfig tlp down' (replace tlp0 with network interface
 	of choice :)

--

--
(Continue reading)

Permalink | Reply | 1 comment |
headers
Nicolas Joly | 3 Mar 2009 09:18
Picon
Picon
Favicon

Re: Netbooting a machine w/ a disk

On Mon, Mar 02, 2009 at 02:42:49PM -0600, Dustin Marquess wrote:
> On Mon, Mar 2, 2009 at 3:22 AM, David Brownlee <abs <at> netbsd.org> wrote:
> 
> Only weird thing that I've noticed is that once I'm in NetBSD, if I
> reboot/halt back to SRM and try to boot again, BOOTP just sits there
> and times out, almost like network connectivity isn't working.  But if
> I 'init', it works again.  Very strange..

Looks like a problem i already encountered (PR/40077) ... I cannot
reboot/halt because pmf_system_shutdown() has disabled the PCI bus
that hold the disks.

Can you try commenting `pmf_system_shutdown()' call in cpu_reboot()
from sys/arch/alpha/alpha/machdep.c and check if that make it work
better.

Thanks.

--

-- 
Nicolas Joly

Biological Software and Databanks.
Institut Pasteur, Paris.
Permalink | Reply | 1 comment |
headers
Dustin Marquess | 3 Mar 2009 21:23
Picon

Re: Netbooting a machine w/ a disk

On Mon, Mar 2, 2009 at 4:58 PM, David Brownlee <abs <at> netbsd.org> wrote:
> On Mon, 2 Mar 2009, Dustin Marquess wrote:
>
>>>        If you set the raid to autoconfigure as a root filesystem
>>>        (raidctl raid0 -A root) then a kernel will automatically
>>>        use it as the root filesystem, at least for a standard
>>>        kernel without a hardcoded root device. I've not tried it
>>>        with a kernel which had hardcoded the root device - let us
>>>        know what happens :)
>>
>> Honestly I went ahead and rebuilt a kernel hardcoded to raid0a to be
>> on the safe side, and it worked great.  I still have the old one
>> backed up however that I could try if you really want me to, but
>> honestly I'm slightly scared :).
>
>        Heh, give is a span next time you need to reboot for any
>        other reason maybe.

Went ahead and gave it a shot.  Didn't work (Yes, I know running mixed
disks isn't optimal :) ):

sd0 at scsibus1 target 0 lun 0: <SEAGATE, ST318406LC, 010A> disk fixed
sd0: 17501 MB, 26302 cyl, 2 head, 681 sec, 512 bytes/sect x 35843670 sectors
sd0: sync (12.50ns offset 63), 16-bit (160.000MB/s) transfers, tagged queueing
sd1 at scsibus1 target 1 lun 0: <FUJITSU, MAS3184NP, 0104> disk fixed
sd1: 17524 MB, 27206 cyl, 2 head, 659 sec, 512 bytes/sect x 35890512 sectors
sd1: sync (6.25ns offset 127), 16-bit (320.000MB/s) transfers, tagged queueing
Kernelized RAIDframe activated
raid0: RAID Level 1
raid0: Components: /dev/sd0a /dev/sd1a
(Continue reading)

Permalink | Reply | 0 comments |
headers
Dustin Marquess | 3 Mar 2009 21:24
Picon

Re: Netbooting a machine w/ a disk

On Tue, Mar 3, 2009 at 2:18 AM, Nicolas Joly <njoly <at> pasteur.fr> wrote:
> On Mon, Mar 02, 2009 at 02:42:49PM -0600, Dustin Marquess wrote:
>> On Mon, Mar 2, 2009 at 3:22 AM, David Brownlee <abs <at> netbsd.org> wrote:
>>
>> Only weird thing that I've noticed is that once I'm in NetBSD, if I
>> reboot/halt back to SRM and try to boot again, BOOTP just sits there
>> and times out, almost like network connectivity isn't working.  But if
>> I 'init', it works again.  Very strange..
>
> Looks like a problem i already encountered (PR/40077) ... I cannot
> reboot/halt because pmf_system_shutdown() has disabled the PCI bus
> that hold the disks.
>
> Can you try commenting `pmf_system_shutdown()' call in cpu_reboot()
> from sys/arch/alpha/alpha/machdep.c and check if that make it work
> better.

Okay, in 5.0RC2, machdep.c:cpu_reboot() calls
kern_subr.c:doshutdownhooks(), which is what calls
pmf_system_shutdown().  Commenting-out pmf_system_shutdown() in
doshutdownhooks() does indeed appear to fix the reboot issue.  I'm not
sure if there are any side effects in doing this vs letting SRM fully
reinitialize everything, but everything at least appears to work
correctly.  It isn't a huge deal for me as I keep serial consoles on
all of my alpha boxen, but I'm sure others would greatly appreciate it
:).

Thanks!
-Dustin
(Continue reading)

Permalink | Reply | 0 comments |
headers
Steve Rikli | 4 Mar 2009 16:05

Free RAM for AlphaServer DS* in Silicon Valley, USA

I've got a few packets of RAM for AlphaServer DS10/20 to give away.

The stuff is 256MB sticks of 200pin Infineon RAM, Sync, CL2, ECC,
Registered, 100MHz.

I know it works in DS10 and DS20 because that's where it was used.
Disclaimer:  it has all been in static bags for a while, so I'm not
100% it is still good.  It was okay when it was pulled, at least.

It is _not_ compatible with the 168pin PWS systems like the 500au
and similar -- sorry.

I have 4 packets of 4 (i.e. 16 total) and all I ask in return is
reimbursement for shipping.  Hopefully there are still some
AlphaServers out there running NetBSD to use this stuff!

Cheers,
sr.
Permalink | Reply | 0 comments |
headers
Havard Eidnes | 10 Mar 2009 09:45
Picon

Re: Package binaries for NetBSD/alpha 3.1 / pkgsrc-2008Q4

Hi,

I've uploaded the results of a bulk rebuild for NetBSD/alpha
3.1 to

   ftp://ftp.NetBSD.org/pub/pkgsrc/packages/NetBSD/alpha/3.1_2008Q4/

A total of 2.5GB of packages were uploaded for a total count of
2718 packages.

The list packages available in new versions or which were not available
earlier is: GraphicsMagick-1.3.5, asterisk-1.2.31, awstats-6.9,
bacula-clientonly-2.4.3nb1, bind-9.4.3pl1, bind-9.5.1pl1, bugzilla-2.22.7,
bugzilla-3.0.8, dbus-1.2.4.2nb2, docker-1.5nb1, dovecot-1.1.10,
drupal-6.9, ejabberd-2.0.3, fftwf-3.2, horde-3.3.3, imp-4.3.3,
libaudiofile-0.2.6nb2, mediawiki-1.13.4, mysql-client-5.0.67nb2,
mysql-server-5.0.67nb1, ntp-4.2.4p6, openssl-0.9.8j,
p5-CatalystX-CRUD-Controller-RHTMLO-0.19, p5-CatalystX-CRUD-YUI-0.012,
p5-MIME-Lite-3.023nb1, p5-Rose-DBx-Garden-0.15nb1,
p5-Rose-DBx-Garden-Catalyst-0.13, p5-Rose-HTML-Objects-0.600nb1,
p5-Rose-HTMLx-Form-Field-Autocomplete-0.02nb1,
p5-Rose-HTMLx-Form-Field-Boolean-0.03nb1,
p5-Rose-HTMLx-Form-Field-Serial-0.002nb1, p5-Rose-HTMLx-Form-Related-0.13,
phppgadmin-4.2.2, postfix-2.5.6, proftpd-1.3.2, py24-django-1.0.2nb1,
rrdtool-1.2.30, rrdtool-1.3.6, squid-2.7.6, squid-3.0.13, sudo-1.7.0,
thunderbird-2.0.0.19nb1, thunderbird-gtk1-2.0.0.19nb1, tor-0.2.0.33,
typo3-4.2.4, xterm-238, and xtide-2.8.3nb2.

The source tree this was built from was updated Feb 11 2009, and
the packages were built on a system running 3.1_STABLE.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jarle Greipsland | 17 Mar 2009 08:47
Picon
Picon

Broken build?

I tried to build a GENERIC kernel from -current sources
yesterday, but it bombed out:
    compile  kern/bcmp.o
cc1: warnings being treated as errors
/usr/src/sys/lib/libkern/../../../common/lib/libc/string/bcmp.c:54: warning: no previous
prototype for 'bcmp'
--- bcmp.o ---
*** [bcmp.o] Error code 1
1 error
nbmake: stopped in /usr/obj/sys/arch/alpha/compile/GENERIC/lib/kern

Known problem, or is the problem at my end?

					-jarle
--

-- 
we all hack on a broken subroutine, a broken subroutine, a broken subroutine...
					-- Kenneth Stailey
Permalink | Reply | 1 comment |
headers
Havard Eidnes | 17 Mar 2009 11:15
Picon

Re: Broken build?

> I tried to build a GENERIC kernel from -current sources
> yesterday, but it bombed out:
>     compile  kern/bcmp.o
> cc1: warnings being treated as errors
> /usr/src/sys/lib/libkern/../../../common/lib/libc/string/bcmp.c:54: warning: no previous
prototype for 'bcmp'
> --- bcmp.o ---
> *** [bcmp.o] Error code 1
> 1 error
> nbmake: stopped in /usr/obj/sys/arch/alpha/compile/GENERIC/lib/kern
>
> Known problem, or is the problem at my end?

Sort-of known problem, also affects a number of other ports: all the
powerpc ports and hp700 as well.  The move of the declarations of the
various b* functions in libkern.h under an #ifdef _STANDALONE appears
to be related, but the fix is not as simple as reverting that single
part of the change (which appear to have been done at the same time
the tree was de-__P'ed).

Regards,

- Håvard
Permalink | Reply | 0 comments |

Gmane