Re: [9fans] (no subject)

Re: [9fans] (no subject)

[YAMANASHI Takeshi]

> Why not generate a proto file for mkfs/mkext to copy the
> minimal set of files for the auth server to your compact
> flash, and then post the proto file on the wiki?

Here is my proto file.  I used this to copy necesary files
into a single dos floppy.  My auth server is two-floppy
(one for 9load & kernel, the other for fs) based standalone
server.
--
386
	init
	bin
		rc
		cat
		echo
		ls
		ps
		date
		ip
			ipconfig
		ndb
			cs
		aux
			listen
		auth
			keyfs
			changeuser
			authsrv
adm
	timezone
		local
	keys
	keys.who
rc
	bin
		service
		service.auth
			authsrv.il566
			authsrv.tcp567
	lib
		rcmain
lib
	namespace
	ndb
		local
		auth
mnt
	keys
	netkeys

[9fans] usb keychain boot?

[Tim Newsham]

Is it possible to run the plan9 boot loader from a USB keychain
(not to boot the entire system from the keychain, just to get
the tftp boot to run)?

Tim N.

Re: [9fans] usb keychain boot?

[Tim Newsham]

similar question -- how does the 9pxeload loader work if being
served from a plan9 server?  If I setup the tftp server to
serve "9pxeload", how does it then serve up a kernel when
9pxeload runs?

Tim N.

Re: [9fans] usb keychain boot?

[Kenji Okamoto]

> similar question -- how does the 9pxeload loader work if being
> served from a plan9 server?  If I setup the tftp server to
> serve "9pxeload", how does it then serve up a kernel when
> 9pxeload runs?

Sure, it boot from PXE setted by existence of /lib/tftpd/9pxeload,
and the client machine has a line of
	bootf=9pxeload
in her part /lib/ndb/local file.

However, I don't know how to set up plan9.ini in this case though.

Kenji

[9fans] local mail question

[Tim Newsham]

The wiki discusses how to configure plan9 to send and receive
mail outside the system.  I am interested in sending mail locally
on the system.  When I use the acme mail client to mail myself
or to mail glenda I get errors:

   Mail from 'newsham'
   to 'newsham'
   failed with error 'Invalid address'.

   saved in /mail/box/newsham/dead.letter

I followed the wiki's directions for creating mailboxes.  I noticed
that this creates files with slightly different permissions than
the glenda box:

  term% ls -l . newsham glenda
  d-rwxrwxr-x M 9 glenda  glenda 0 May 22  2002 ./glenda
  d-rwxrwxr-x M 9 newsham upas   0 Jul 31 16:05 ./newsham
  -lrw-rw-rw- M 9 newsham upas   0 Sep 25  2002 newsham/L.mbox
  --rw-rw---- M 9 newsham upas 167 Jul 31 16:07 newsham/dead.letter
  alrw--w--w- M 9 newsham upas   0 Sep 25  2002 newsham/mbox
  -lrw-rw-rw- M 9 glenda  glenda     0 Apr 13  2002 glenda/L.mbox
  a-rw-rw--w- M 9 glenda  glenda 10699 Apr 22  2002 glenda/mbox

Which is right?  What needs to be done to allow sending to local
boxes?

Tim N.

[9fans] ndb question

[Tim Newsham]

This doesnt look right.  I edit /rc/bin/termrc, I get rid of the
"cpu=XXX" line.  I edit /lib/ndb/local and setup a default cpu setting
and verify with:

   ndb/ipquery ip <my ip> cpu     (result:  cpu=XXX)

but when I boot the system the cpu variable is unset.  How is the
cpu setting in ndb used, if at all?  I know I can set this in
termrc, but ndb seems more flexible.  I could always set it
from an ipquery, but this seems hacky.  I'm trying to set things
up "properly" here.

Tim N.

Re: [9fans] local mail question

[geoff]

Each user who wants to receive mail needs to run 'mail -c'.  This
could be done in /sys/lib/newuser, though it isn't in the stock one.

I suspect that it's a rewrite problem.  Try

	mail '-dd#' newsham

and see if the problem becomes obvious.  `Invalid address' usually
means that upas fell off the end of your rewrite file without a match.
You ought to have a line like this to deliver local mail:

	# local mail
	local!([^!]+)		>> /mail/box/\1/mbox

Order matters, so be careful.  If in doubt, look at
/mail/lib/rewrite.* as examples.

Re: [9fans] ndb question

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 471 bytes --]

try setting  cpu='$cpu' in termrc, with the cpu=... entry in ndb.

the auth=, cpu= ... names in ndb are typically used when cs (via dial) is asked to
look for $X for some X as part of a dial string (eg, net!$auth!fsauth).
since the cpu command puts the value of the environment variable cpu
in a dial string, setting it to the literal '$cpu' puts that in the dial string,
where cs will translate it by looking in ndb.

see the description of metanames in ndb(6)

[-- Attachment #2: Type: message/rfc822, Size: 2934 bytes --]

From: Tim Newsham <newsham@lava.net>
To: Fans of the OS Plan 9 from Bell Labs <9fans@cse.psu.edu>
Subject: [9fans] ndb question
Date: Sat, 31 Jul 2004 16:59:30 -1000 (HST)
Message-ID: <Pine.BSI.4.58.0407311653340.9843@malasada.lava.net>

This doesnt look right.  I edit /rc/bin/termrc, I get rid of the
"cpu=XXX" line.  I edit /lib/ndb/local and setup a default cpu setting
and verify with:

   ndb/ipquery ip <my ip> cpu     (result:  cpu=XXX)

but when I boot the system the cpu variable is unset.  How is the
cpu setting in ndb used, if at all?  I know I can set this in
termrc, but ndb seems more flexible.  I could always set it
from an ipquery, but this seems hacky.  I'm trying to set things
up "properly" here.

Tim N.