| Author |
having trouble booting Ultra 30 over network
|
|
| Jeffrey Moncrieff 2004-12-09, 8:49 pm |
| Hello
I am trying to boot my urlta 30 when I do a boot net - install all I
get is a count up to 29200 and the beeps I know the mb is good
beacuse it boot form hard disk. my boot server is linux based. using
tftp and rarpd the sparc has openboot 3.11
please cc me your comments
thanks jeff
jeff_moncrieff6@sympatico.ca
| |
| Bernd.Schemmer 2004-12-10, 5:55 pm |
| Jeffrey Moncrieff wrote:
> Hello
>
> I am trying to boot my urlta 30 when I do a boot net - install all I
> get is a count up to 29200 and the beeps I know the mb is good
> beacuse it boot form hard disk. my boot server is linux based. using
> tftp and rarpd the sparc has openboot 3.11
What Linux version?
If I remember correct net booting Solaris is only possible with newer
Linux Versions due to an incompability of tftpd/tftp between Linux and
Solaris
regards
Bernd
>
>
> please cc me your comments
> thanks jeff
> jeff_moncrieff6@sympatico.ca
| |
| Fredrik Lundholm 2004-12-10, 5:55 pm |
| In article <e7be4630.0412091816.779b2f5b@posting.google.com>,
Jeffrey Moncrieff <jeff_moncrieff6@sympatico.ca> wrote:
>Hello
>
>I am trying to boot my urlta 30 when I do a boot net - install all I
>get is a count up to 29200 and the beeps I know the mb is good
>beacuse it boot form hard disk. my boot server is linux based. using
>tftp and rarpd the sparc has openboot 3.11
The latest OBP for Ultra 30 can use dhcp for booting which might help.
Also you are not mentioning a bootparams-daemon which should also
be neccessary for a correct jumpstart server.
/wfr
Fredrik
--
Fredrik Lundholm
dol @ ce.chalmers.se
| |
| Darren Dunham 2004-12-10, 5:55 pm |
| In comp.unix.solaris Jeffrey Moncrieff <jeff_moncrieff6@sympatico.ca> wrote:
> Hello
> I am trying to boot my urlta 30 when I do a boot net - install all I
> get is a count up to 29200 and the beeps I know the mb is good
> beacuse it boot form hard disk. my boot server is linux based. using
> tftp and rarpd the sparc has openboot 3.11
After the count, the TFTP process is done. The client should be
attempting to perform rpc.bootparamd calls to get information, followed
by NFS mounts. Snoop/tcpdump the wire to see what it's doing.
--
Darren Dunham ddunham@taos.com
Senior Technical Consultant TAOS http://www.taos.com/
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >
| |
| Darren Dunham 2004-12-10, 5:55 pm |
| In comp.unix.solaris Bernd.Schemmer <bnsmb@online.de> wrote:
[vbcol=seagreen]
> What Linux version?
> If I remember correct net booting Solaris is only possible with newer
> Linux Versions due to an incompability of tftpd/tftp between Linux and
> Solaris
I've heard of that before, but I've never seen it. Way back in the 0.9
versions of Linux, we had a boot server that worked just fine at the
rarp/tftp/rpc.bootparamd levels. It was always the NFS stuff that
broke.
--
Darren Dunham ddunham@taos.com
Senior Technical Consultant TAOS http://www.taos.com/
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. >
| |
| Fredrik Lundholm 2004-12-16, 7:45 pm |
| In article <Dqpud.40970$6q2.39257@newssvr14.news.prodigy.com>,
Darren Dunham <ddunham@redwood.taos.com> wrote:
>In comp.unix.solaris Bernd.Schemmer <bnsmb@online.de> wrote:
[vbcol=seagreen]
>
>I've heard of that before, but I've never seen it. Way back in the 0.9
>versions of Linux, we had a boot server that worked just fine at the
>rarp/tftp/rpc.bootparamd levels. It was always the NFS stuff that
>broke.
Actually really old linux worked, it was when some clever person made
a check in tftpd "for security reasons" so that an absolute pathname
was always reqiured. If you look what a Sun requests it was in a "bad"
format. The easy workaround for redhat was to eliminate the check in the
SRPM and use that.
/wfr
Fredrik
--
Fredrik Lundholm
dol @ ce.chalmers.se
| |
| Helmut Kreft 2004-12-16, 7:45 pm |
| On 9 Dec 2004 18:16:08 -0800, Jeffrey Moncrieff wrote:
>
> I am trying to boot my urlta 30 when I do a boot net - install all I
> get is a count up to 29200 and the beeps I know the mb is good
> beacuse it boot form hard disk.
This count really only tells you that rarpd and tftp are probably working
as expected. If you see nothing after that, I think something is seriously
wrong. There should at least be the message "Requesting Internet address for
0:3:ba:5b:b9:f7" and a whirling busy indicator. If this is not there you
probably have an issue with the tftpboot (file corrupt, wrong file, etc.)
If the whirling thing just keeps whirling your rpc.bootparamd might have a
problem (not running, not reachable, ...).
> my boot server is linux based. using
> tftp and rarpd the sparc has openboot 3.11
>
My last try with that was a stock SuSE 8.1 vs. Solaris 8. It took hours
of analyzing tcpdumps to get this working. I suggest you abadon this idea
if at all possible.
The major problems were:
- In my network some bootparam broadcast requests were sent to x.y.255.255
which would have been correct if this net wasn't subnetted. The Solaris
rpc.bootparamd does not care and answers anyway. With Linux you will
see the boot fail at some point.
- NFSv3 at least on my SuSE featured an implementation for ACLs which
Solaris did not like. Exporting everything rw helped.
Helmut
--
Half the people you know are below average.
| |
| David Douthitt 2004-12-30, 2:47 am |
| Darren Dunham wrote:
> In comp.unix.solaris Bernd.Schemmer <bnsmb@online.de> wrote:
>
>
>
>
> I've heard of that before, but I've never seen it. Way back in the 0.9
> versions of Linux, we had a boot server that worked just fine at the
> rarp/tftp/rpc.bootparamd levels. It was always the NFS stuff that
> broke.
I experienced the same symptoms as the original poster with an Ultra
Enterprise 1 with a CentOS 3 (Red Hat Advanced Server 3) Linux installation.
When I switched the Install Server to a Solaris 8 x86 box, it worked
fine. Note that I did not move the (Solaris) Boot Server - this
remained on the Linux box. Since NFS problems have been reported
before, I suspect this was the problem again. I did see NFS traffic
going across the wire when snooping with tcpdump, so this sounded logical.
|
|
|
|