[Fwd: [Bug translator/6565] network compile server, stage 1]

[Dave Brolley <brolley at redhat dot com>]

I just pushed an updated prototype stap client/server. Details below 
from bz6565.

In order to set up a server, one may now simply run stap-serverd <port>. 
The server will automatically advertise its presense on the network and 
will use <port> and <port>+1 (see issues below). Multiple servers may be 
running concurrently using different ports (I'm working on automating 
this). The default ports are 65000 and 65001. stap-serverd may be killed 
by sending it SIGINT.

To use the server one simply runs stap-client on a compatible (or the 
same) machine on the local network. stap-client takes the same arguments 
as stap and will automatically find a compatible server (see issues 
below) if one is available.

My current issues with the implementation are:

o I still need to finish defining what the client should be looking for 
when looking for a compatible server. There has been some loose 
discussion, but I need to nail this down.  Currently only `uname -r` is 
checked.

o How does the client establish that a given server is trusted.  I've 
read Roland's remarks on the subject of client/server security and will 
be reading them again to see if any of it will sink in.

o In order to get around the problem of tcp connections being closed 
after a file is sent, the current implementation uses one port for 
synchronization (which remains connected during the entire transaction) 
and another for file transfer (which is opened and closed as each file 
is transferred). My concern is that this presents an opportunity for 
this second port to be hijacked by an imposter either in order to 
present the server with an alternate request and/or to steal the 
server's response. Any ideas on how both the client and server can 
determine that the same client/server is using the second port would be 
much appreciated. Perhaps thes time that this whole thing was written in 
a higher level language has come. Perhaps the answer lies in Roland's 
comments.

Dave

-------- Original Message --------
Subject: 	[Bug translator/6565] network compile server, stage 1
Date: 	3 Jul 2008 17:32:50 -0000
From: 	brolley at redhat dot com <sourceware-bugzilla@sourceware.org>
Reply-To: 	sourceware-bugzilla@sourceware.org
To: 	brolley@redhat.com
References: 	<20080526195706.6565.fche@redhat.com>



------- Additional Comments From brolley at redhat dot com  2008-07-03 17:32 -------
I just pushed a new prototype version of the stap client/server. It addresses
concerns raised about the initial version, including:

o Server no longer managed by xinetd. Instead a new stap-serverd script can be
started by anyone. Multiple stap-serverd instances can run simultaneously on
separate (specified) ports. stap-serverd can be killed by sending it a SIGINT
signal.

o The /etc/avahi/services config file is now obsolete. stap-serverd now uses
avahi-publish-service to advertise its own presence on the network.

o Communication protocol has been simplified so that no trusted information is
used to synchronize the flow of communication. rsync is no longer used to
transer files. ssh access to the server machine is no longer required. nc is
used for all client/server communication.

o Both the client and the server do their own examination of the files passed in
order to verify their appropriateness.



--


http://sourceware.org/bugzilla/show_bug.cgi?id=6565

------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.