From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <083ab2fec3a16915da8da52b07c2b81c@plan9.ucalgary.ca>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] Plan9 and resource discovery
Date: Tue, 23 Nov 2004 21:22:09 -0700
From: andrey mirtchovski <mirtchov@cpsc.ucalgary.ca>
In-Reply-To: <20041124040536.GA17538@submarine>
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Topicbox-Message-UUID: 0ae3d812-eace-11e9-9e20-41e7f4b1d025


> I guess my question is -- is there any research going on
> on how Plan9's philosophy could lend itself to this problem?
> Any ideas ? Implementations ? Anything ? :-)

i have a prototype for a resource discovery fs.  each server that
wants to participate runs a small fs which gathers ldap-like info
about the local system.  each server posts the root of that file
system to a main aggregate which is listed in ndb for the local ipnet
(domain).  a user mounts either a the node itself or all of them via
the aggregate.  aggregates themselves can be aggregated for
cross-domain resource discovery.  run 'ls' and you find out which
nodes are alive, 'cat */load | sort' and you get the one that's least
busy...

it's not anything big, but then again Plan 9 doesn't easily lend
itself to big and complex solutions for trivial problems.

andrey