gmane.os.netbsd.general

Is there a BSD version of tracepath?

Fredrik Pettai — 2012-05-25T12:13:01

Hi,

I was looking for a tool for discovering the MTU on a routed path (with several hops), and "tracepath" seems to be the right tool for that. However, tracepath is part of (Linux) iputils which seem to be just Linux centric. 

Is someone aware of and similar tool for *BSD systems?

Re,
/P

Re: Booting a NetBSD installer from pxelinux

Roger Pau Monné — 2012-05-25T09:10:02

Thanks, passing the io address solved the issue, now NetBSD correctly
finds the com port.

Re: Booting a NetBSD installer from pxelinux

Manuel Bouyer — 2012-05-25T07:54:57

You can installboot -e (on a NetBSD box, e.g:
installboot -e -o console=com0,speed=9600,ioaddr=0x3f8 netbsdpxe.0) the
NetBSD PXE boot loader to set the console parameters.
This way it will automatically start on the serial port, you don't need
to type 'consdev comx'.

Re: Booting a NetBSD installer from pxelinux

Roger Pau Monné — 2012-05-25T07:47:21

My main problem is that I would like to realize the install from the
serial port, but since "consdev comx" doesn't seem to work I don't
know what other options I have.

I've also realized that when booting from VGA with the 6.0BETA install
CD the install process gets stuck after something like:

boot device:
root device:

And there's no output behind this point (although I see the CD gets
recognized by the kernel). That's why I wanted to use the serial
console, so I can paste the output of the boot process and hopefully
find the error.

Thanks, Roger.

Re: sbc better than soekris?

James K. Lowden — 2012-05-24T22:49:12

On Sun, 20 May 2012 17:28:57 -0400
"James K. Lowden" <jklowden< at >schemamania.org> wrote:


Many thanks to everyone for your suggestions.  When my 4801 was first
installed, I put in a 256 MB CF card and 30 GB laptop HD.  Now in 2012
a 16 GB CF Card is $30.  That certainly seems like a the best
choice: good fast and cheap.  

--jkl

Re: Booting a NetBSD installer from pxelinux

Manuel Bouyer — 2012-05-24T17:16:44

I've not tried 'consdev comx', but installboot -e does work for me.
I've found that on some systems the BIOS doesn't properly export com
parameters, and installboot also needs the I/O port and speed ...

Re: Booting a NetBSD installer from pxelinux

Roger Pau Monné — 2012-05-24T17:08:15

Thanks!, this is really and advance. I can get to the NetBSD
bootloader, and using boot tftp:path/to/netbsd-INSTALL.gz finds and
loads the kernel, but it's tedious to write boot tfp:path/to/kernel,
is there some way to pass a config file as an argument to netbsdpxe.0?
That would be great.

Another problem that I've found (both from the BETA6 CD or the PXE
bootloader) is that "consdev comx" doesn't seem to work for me. I've
tried all the comX options, but the kernel always throws it's output
to the VGA port, any thoughts on this one? This is a Dell R310 by the
way.

This is the output I get, but I think it won't be of much help:

boot: netbsd


Press return to boot now, any other key for boot menu
booting netbsd - starting in 0 seconds.
type "?" or "help" for help.


PXE BIOS Version 2.1
Using PCI device at bus 2 device 0 function 0
Ethernet address 84:2b:2b:69:bd:cf
11628440+5725352+620280=0x1224a70

Thanks, Roger.

Re: Booting a NetBSD installer from pxelinux

Manuel Bouyer — 2012-05-24T16:05:44

FWIW, I load the netbsd bootloader from pxelinux, which then loads the
NetBSD kernel:
label netbsd
  kernel netbsdpxe.0

netbsdpxe.0 being /usr/mdec/pxeboot_ia32.bin
This then allows me to load the kernel I want, load ramdisk modules, etc ...

Booting a NetBSD installer from pxelinux

Roger Pau Monné — 2012-05-24T14:56:07

Hello,

I've been trying to boot a netbsd INSTALL kernel from pxelinux, but so
far I haven't got much success. I've tried with 6.0BETA amd64 INSTALL
kernels and it seems to be compiled without MULTIBOOT options, which
makes them useless with pxelinux. I've tried adding "options
MULTIBOOT" to the sys/arch/amd64/conf/INSTALL conf file, but the
resulting kernel is still not booting with pxelinux. I'm using the
following pxelinux configuration:

label netbsd6-install
        kernel mboot.c32
        append rogerpau/netbsd-INSTALL

Which gives "Fatal: not a multiboot kernel." error when using amd64
versions, but with i386 I've got a little bit more success, I'm able
to boot the install kernel, that seems to come with multiboot
activated by default on config file, but since I'm on a serial console
I'm not able to see any output, and the machine reboots shortly after
booting the kernel, to get serial console output I've tried passing
"consdev com1", but I'm not really sure this is the correct syntax to
use with pxe

Re: GPT and raid seems ok, missing disks awkward

Petar Bogdanovic — 2012-05-23T14:15:09

Sorry that's not true. It would only be 64 if the ffs partition of your
second gpt would also start at 64.  The first gpt/mbr layout doesn't
matter in this case.

Re: GPT and raid seems ok, missing disks awkward

Petar Bogdanovic — 2012-05-23T14:09:32

Yes, boot knows how to deal with disklabel < at > raidframe so as long as you
stick with it you won't need that hack even if your raidframe is < at > gpt < at >
disk.

(sorry for the silly < at >-notation)

Re: GPT and raid seems ok, missing disks awkward

Greg Troxel — 2012-05-23T14:01:12

  If you intend to boot from this (ffs < at > gpt < at > raidframe < at > gpt < at > disk),
  you'll probably need this ugly duckling:

     -       bios_sector += RF_PROTECTED_SECTORS;
     +       bios_sector += RF_PROTECTED_SECTORS+34;

Interesting point.  What I have done is

  gpt on disk, with raidframe partition at 64
  in raid, regular disklabel
    raid0a starting at 0

I think for 2T (vs 2.5T or bigger) I don't really need gpt, but I like
having a clear label on the disk show up in dmesg when it attaches.

So I think boot will only need to skip the 64 RF sectors after first
finding the raid partition from parsing the gpt label.  Does that sound
right?

Re: GPT and raid seems ok, missing disks awkward

Petar Bogdanovic — 2012-05-23T13:24:38


If you intend to boot from this (ffs < at > gpt < at > raidframe < at > gpt < at > disk),
you'll probably need this ugly duckling:

    Index: bootxx/boot1.c                             
    ===================================================================
    RCS file: /cvsroot/src/sys/arch/i386/stand/bootxx/boot1.c,v        
    retrieving revision 1.20                                   
    diff -u -r1.20 boot1.c  
    --- bootxx/boot1.c      6 Jan 2011 01:08:48 -0000       1.20
    +++ bootxx/boot1.c      30 Mar 2012 13:58:33 -0000          
    < at >< at > -86,7 +86,7 < at >< at >                                 
             * (Maybe this should only be done if the filesystem
             * magic number is absent.)
             */
    -       bios_sector += RF_PROTECTED_SECTORS;
    +       bios_sector += RF_PROTECTED_SECTORS+34;
            fd = ob();
            if (fd != -1)
                    goto done;
    Index: lib/biosdisk.c     
    ===================================================================
    RCS file: /cvsroot/src/

GPT and raid seems ok, missing disks awkward

Greg Troxel — 2012-05-23T12:44:10

I ended up figuring out my issues, but I'm thinking this note may be
useful to others.

summary:
  Does raidframe on dkN via gpt work? (yes, it seems to)
  Does the "absent" keyword in raidctl still work? (yes, but with -C)

My system is up-to-date netbsd-5/i386.

I just got two 2T SATA drives, intending to migrate a system from 2 *
400G RAID-1 to a RAID-1 pair with the new drives.  (Not relevant, I
think: I got one Seagate and one Hitachi drive, both 7200 rpm, to reduce
the chance of correlated failures.)

I put the Seagate drive in an Aluratek docking enclosure, and it shows
up as sd4 and seems to work.

I used gpt create, gpt add, gpt label, and addwedge, and can access dk0.
It looks fine (yes, I started at 64 not 34 due to 4K sectors which I
probably have):

       start        size  index  contents
           0           1         PMBR
           1           1         Pri GPT header
           2          32         Pri GPT table
          34          30         
          64  3907029071      1  GPT par

How to install the i386 boot when upgrading from 5.1 to 6.0?

Ji YongGang — 2012-05-23T05:16:50

Hi all,

Recently I want to upgrade my thinkpad x60 from netbsd/i386 5.1 to 6.0
branch, so I cvs update the /usr/src

$ cvs up -dP -r netbsd-6

and following the instruction:

http://www.netbsd.org/docs/guide/en/chap-updating.html#updating-summary

After I build and replace the GENERIC kernel with new one, the
reboot step cannot succeeded.

The screen only display the first line message and reboot immediately.

I knew the new netbsd-6 kernel using modules under i386, so I try the
MONOLITHIC kernel, but it failed boot, too.

Then I copy the fresh 6.0 destdir.i386/usr/mdec/boot to /boot,
destdir.i36/stand/i386/6.0 to /stand/i386. Boot the system, this time
the screen show more one line:

Loading /stand/i386/6.0/modules/ffs/ffs.kmod

and reboot again.

Add -vx boot options, no verbose and debug messages appeared.

I also try the same upgrading procedure on another PC box, it has
netbsd/i386 5.1 installed, and boot new kernel encounter the same reboot
problem.

The bootloader on these systems is GRUB 0.97. 

Did

Re: Can't mount iso image on virtual node

Michael van Elst — 2012-05-21T20:51:19



You could install the cdrtools package. It contains the programs
isoinfo to dump the directory from an ISO9660 image and isovfy to
verify an ISO9660 image.

Re: Can't mount iso image on virtual node

Mayuresh — 2012-05-21T15:41:56

I took some risk and went ahead with multission write with above iso
(which I could not verify due to said issue).

The write was successful. I also verified the contents on CD after write.

So I suspect, the iso created for a multisession write won't be mountable
for viewing its contents.  Can anyone confirm?

Mayuresh.

Re: Check condition on CDB (Was: Can't mount iso image on virtual node)

Mayuresh — 2012-05-21T15:39:05

It is observed that on rebooting the error goes away. It is unclear when
it sets in. However once it sets in, it doesn't go away until next reboot.

Mayuresh.

Re: 5.1_STABLE/i386 panic after recent pull-ups

Manuel Bouyer — 2012-05-21T13:50:42

Please send-pr this

Re: sbc better than soekris?

Petar Bogdanovic — 2012-05-21T12:19:16

My personal rule of thumb is: as long as it doesnt move, it will very
likely last for a long time.

So why don't you just replace the Disk and pax your data?  You'll get
several hours(/days) to invest elsewhere for free. :)



If you trust their Specs (MTBF), yes.


Also you may want to take a look at the new Soekris net6501 (Intel Atom)
boards.  If my two Alix boards weren't running that well, I'd go for one
of those, especially since they have several Gig-E Ports.

http://soekris.com/products/net6501.html

Re: sbc better than soekris?

Greg A. Woods — 2012-05-21T03:24:58

Subject: sbc better than soekris?

I'd suggest updating to a newer Soekris 5501 model, but unfortunately
they're still rather expensive.

There are versions of PC Engines ALIX boards which are substantially
similar to the newer Soekris 5501 and they are sometimes available on
E-bay for more reasonable prices.  I don't think they are quite as nice
though, but functionally they do much the same job (you only get two, or
three, Ethernet ports on the most common models though).

You may find something decent in the SBC SoC-based ARM or MIPS world
that works well with NetBSD, but finding such devices with multiple
proper separate Ethernet ports is difficult.  Many now have wonky
special switch chips, and work is just now being done to get FreeBSD
drivers for them (and hopefully they can be pulled into NetBSD soon).

I ended up buying a fully kitted out Soekris 5501-70 for myself.

As for storage for such devices, well that's yet another good question.

I picked up a 16GB Compact FASH card for my Soekris machine a