another CPU

another CPU

[thinktankworkspaces]

Rather than doing another 9front install to native install. Can't I just
use the 9front.iso and pick tcp from the menu give it

user[glenda]:
fs address is? 192.168.1.172
auth address is? [192.168.1.172]
srv: dial tcp!192.168.1.172!564: connection refused
mount: can't open #s/boot: '#s/boot' file does not exist
mount -c #s/boot /root: mount 176: open

I kind of feel like i'm missing some info here. Do I just put another
entry in my cpu/auth 

sys=cpufoo dom=foo.yoda ether=123456789 ip=192.168.1.100
	bootf=/386/9bootpxe


I'm guessing it won't get to it because booting from the iso it doesn't
know anything about the network because no network stack has been 
setup? This is why you kind of need a full install so you can setup a network stack and give it the same address you specify 
on the cpu/auth as another cpu?

I kind of feel like more is also missing on the fs side like a mount
point or something I must have skipped. my CPU/Auth is on Qemu 
and I thought rather than change what I have I could create another 
cpu for speed/performance. And learn at the same time on how to get
another node on the grid?

Re: [9front] another CPU

[hiro]

make sure you're not running 2 dhcp servers.
you need to setup 9front as cpu server, fileserver and dhcp server.
plan9 dhcp server will give above ip=192.168.1.100 to any PC that
requests an IP if they have 123456789 ethernet address.
i don't see any ip subnet, network info above, so perhaps the local
routes are missing, too.

Re: [9front] another CPU

[thinktankworkspaces]

[-- Attachment #1: Type: text/plain, Size: 2028 bytes --]

So I have it working. But maybe i'm missing something. Much
of the documentation talks about

cd /cfg; mkdir $sysname; dircp example $sysname

This seems old and deprecated. I'm not really sure what we
are supposed to have in /cfg/$sysname

I tried copying cpurc and termrc and booting with each of them
the results were strange and buggy. However I made not changes
to those files as well. 

The booting part was the tricky part. I did have two dhcp server
running. One for my internet and one for the CPU/Auth.

ipnet=yoda ip=192.168.1.0 ipmask=255.255.255.0
	ipgw=192.168.1.1
	dns=8.8.8.8
	auth=192.168.1.172
	dnsdom=yoda
	authdom=yoda
	fs=192.168.1.1
	cpu=foo
	smtp=foo

sys=localhost dom=localhost ip=127.0.0.1
sys=foo dom=foo.yoda ether=525400123336 ip=192.168.1.172
sys=cirno dom=cirno.thinktank ether=080027f9ec98 ip=172.27.0.67
#sys=cpubar dom=foo.yoda ether=525400124242 ip=192.168.1.42 bootf=/386/bootx64.efi
sys=cpubar dom=foo.yoda ether=525400124242 ip=192.168.1.42

I did have to make a change to cpubar below. evidently bootx64.efi does not work. 

When I do manage to boot cpubar it takes the name of foo. However
the file system is different as expected. 

netaudit
checking this host's tuple:
	ip=192.168.1.42 looks ok
	dom=foo.yoda looks ok
	no ether entry
checking the network tuple:
	we are not in an ipnet, so looking for entries in host tuple only
	ipgw=192.168.1.1 looks ok
	dns=8.8.8.8 looks ok
	auth=192.168.1.172 looks ok
	fs=192.168.1.1 looks ok
checking auth server configuration:
	we are not the auth server 192.168.1.172
	if this is a mistake, set auth=foo or auth=foo.yoda
	run auth/debug to test the auth server

But i don't see a default mount for the cpu to copy data over 
/n/other/ is the same for both 

When I look back the only thing I did or the only change I made is one
line to /lib/ndb/local and an install on another system. It seems the less
I do the more results as opposed to following a lot of the old 
documentation. But I still seem to be missing a few things. 

Any thoughts?

[-- Attachment #2: Type: message/rfc822, Size: 5411 bytes --]

From: hiro <23hiro@gmail.com>
To: 9front@9front.org
Subject: Re: [9front] another CPU
Date: Mon, 3 Aug 2020 08:47:45 +0200
Message-ID: <CAFSF3XPoMzizmFY7Uss=mYWHFMkmxSqjri3MjqHyF5iK3U90JQ@mail.gmail.com>

make sure you're not running 2 dhcp servers.
you need to setup 9front as cpu server, fileserver and dhcp server.
plan9 dhcp server will give above ip=192.168.1.100 to any PC that
requests an IP if they have 123456789 ethernet address.
i don't see any ip subnet, network info above, so perhaps the local
routes are missing, too.

Re: [9front] another CPU

[Ethan Gardener]

On Sat, Aug 8, 2020, at 6:44 AM, thinktankworkspaces@gmail.com wrote:
> So I have it working. But maybe i'm missing something. Much
> of the documentation talks about
> 
> cd /cfg; mkdir $sysname; dircp example $sysname
> 
> This seems old and deprecated. I'm not really sure what we
> are supposed to have in /cfg/$sysname

It's optional. (The documentation should say so.) If you read /rc/bin/^(term cpu)^rc you'll see files in /cfg/$sysname are sourced only if they exist.

Re: [9front] another CPU

[thinktankworkspaces]

[-- Attachment #1: Type: text/plain, Size: 96 bytes --]

What about changing the extension from cpurc to cpurc.local. I was curious about that as well. 

[-- Attachment #2: Type: message/rfc822, Size: 6597 bytes --]

From: "Ethan Gardener" <eekee57@fastmail.fm>
To: 9front@9front.org
Subject: Re: [9front] another CPU
Date: Sat, 08 Aug 2020 09:48:51 +0100
Message-ID: <209c852d-7353-4c6d-b91a-81603faea9d0@www.fastmail.com>

On Sat, Aug 8, 2020, at 6:44 AM, thinktankworkspaces@gmail.com wrote:
> So I have it working. But maybe i'm missing something. Much
> of the documentation talks about
> 
> cd /cfg; mkdir $sysname; dircp example $sysname
> 
> This seems old and deprecated. I'm not really sure what we
> are supposed to have in /cfg/$sysname

It's optional. (The documentation should say so.) If you read /rc/bin/^(term cpu)^rc you'll see files in /cfg/$sysname are sourced only if they exist.

Re: [9front] another CPU

[thinktankworkspaces]

[-- Attachment #1: Type: text/plain, Size: 915 bytes --]

Okay so the clarification is as follows: 

/rc/bin/cpurc.local only pertains to the intial system that
booted such as the first system which is CPU/Auth. 
However you don't really need that file. Its' just a diff of
/rc/bin/cpurc. So really its also optional

The /cfg/$sysname/cpurc is also optional and no cpurc.local
should exist in that directory. It only pertains to machines
that will boot off the initial server (CPU/Auth). And the
file should be a minimum file of things needed such as
auth/secstore or maybe auth/httpd if you wanted to dedicated
the new cpu to just service httpd as a service right?

As far as a mount point to copy data back and forth. No mount
point exists you still have to mout it directly using commands such as having a listener on the server

server% aux/listen1 -tv tcp!*!9999 /bin/exportfs -R -r /

client% srv tcp!192.168.1.172!9999 servrer /n/server

something like that right?

[-- Attachment #2: Type: message/rfc822, Size: 6597 bytes --]

From: "Ethan Gardener" <eekee57@fastmail.fm>
To: 9front@9front.org
Subject: Re: [9front] another CPU
Date: Sat, 08 Aug 2020 09:48:51 +0100
Message-ID: <209c852d-7353-4c6d-b91a-81603faea9d0@www.fastmail.com>

On Sat, Aug 8, 2020, at 6:44 AM, thinktankworkspaces@gmail.com wrote:
> So I have it working. But maybe i'm missing something. Much
> of the documentation talks about
> 
> cd /cfg; mkdir $sysname; dircp example $sysname
> 
> This seems old and deprecated. I'm not really sure what we
> are supposed to have in /cfg/$sysname

It's optional. (The documentation should say so.) If you read /rc/bin/^(term cpu)^rc you'll see files in /cfg/$sysname are sourced only if they exist.

Re: [9front] another CPU

[Romano]

> So I have it working. But maybe i'm missing something. Much
> of the documentation talks about
> 
> cd /cfg; mkdir $sysname; dircp example $sysname
> 
> This seems old and deprecated. I'm not really sure what we
> are supposed to have in /cfg/$sysname
> 
> I tried copying cpurc and termrc and booting with each of them
> the results were strange and buggy. However I made not changes
> to those files as well. 
> ...
> 
> Any thoughts?

I don't think you want to copy cpurc(8) to /cfg/$sysname/ : see the
man pages for init(8) and cpurc(8): cpurc is started by init, and then
/cfg/$sysname/cpurc runs.  So what you're effectively doing, if I am
reading you correctly, is running cpurc(8) twice. This is all I have in my config:

cpu% cat /cfg/$sysname/cpurc
auth/secstored
cpu%