[evla-sw-discuss] An "In Use" flag for the MIB interface

[Kevin Ryan]

Okay, I understand; since the servlet is running on a 'proxy server'  
to the MIB, the MIB will always see the same IP address (the proxy's)  
in the UDP packet.  I forgot that it was getting the 'lock/key' from  
the packet-header.

I guess then the question becomes - should we think about how this  
will scale for the overall system?  Would it be better to supply the  
lock at a higher level than the packet header (i.e. as an explicit  
message-content parameter) so that the ultimate end user is  
identified?  Just a thought.

Kevin


On Aug 15, 2006, at 2:55 PM, Hichem Ben Frej wrote:

> Kevin,
>
> Thanks for you input.
>
> As proposed, 2 or more processes running on the same host will have
> equal access rights. Any MIB locked by one process is accessible  
> for all
> the other processes since, as you mentioned, the share the same IP  
> address.
>
> We referred to the case of servlet because the "real user" in this  
> case
> is not the servlet but the user that is connected to the servlet.
> Lets say that a servlet is running on a server S. User A connects to S
> through the browser. A sends a lock command, if available. The IP
> address used for the lock is the IP address of the machine S on which
> the servlet is running. Let say user B connects to the same servlet
> running on S.  User B has unlimited access to the MIB locked by user A
> since all the commands sent, through server S, have the right password
> ie. the same IP address. It is similar to the case mentioned above.
>
> Thanks,
>
> Hichem
>
> Kevin Ryan wrote:
>> Hichem,
>>
>> I think this is an excellent implementation of the lock - it is nice
>> and simple.
>>
>> Are there situations where two different processes on the same client
>> would conflict (they would have the same IP address)?
>>
>> 'Left on locks' should simply be handled like they are in the real
>> world; by policy -- "Anyone neglecting to remove his lock shall be
>> punished by not more than 6 months of attending code-walkthroughs"
>> It would be complicated to implement software that could prevent  
>> left-
>> on-locks.
>>
>> Also, I don't think I understand what you mean by the servlet/applet
>> thing not working.  Servlets, at least, should work because they have
>> knowledge of the remote client's IP address.  And I would think
>> Applets could send the client's IP back to the remote server (the  
>> MIB).
>>
>> Or am I misunderstanding something?
>>
>> Kevin
>>
>>
>> On Aug 15, 2006, at 12:04 PM, Bill Sahr wrote:
>>
>>
>>> Any application that locks through a servlet or applet is not
>>> guaranteed
>>> to have exclusive use of a lock because all other applications going
>>> through the same servlet or application will be seen as having  
>>> set the
>>> lock, i.e., will have the same originator IP address (the password).
>>>
>>
>> _______________________________________________
>> evla-sw-discuss mailing list
>> evla-sw-discuss at listmgr.cv.nrao.edu
>> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss
>>
>
>
> -- 
> ______________________________________________________________________ 
> ___
>
> Hichem Ben Frej - hbenfrej at aoc.nrao.edu
>
> NRAO Array Operations Center       Phone:  +1-505-835-7292
> P.O. Box 0 (1003 Lopezville Rd)
> Socorro NM 87801
>
>
>
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss