[evla-sw-discuss] MIB Data Port Spec revisions

[Bill Sahr]

Barry,

Re UDP & fragmentation.  I will ask James.  As a partial answer
I incude his reply (of 10/29/2002) when queried on the issue
of UDP fragmentation w.r.t. the service port.
--------------------------------------------------------------------
On 10/29/2002, James Robnett wrote:

    It should be reassembled just fine.   If one of the fragments
is dropped  then the recieving end will be ableto tell since each
packet has in the headers the total payload size and the start point
of the next fragment.   If it detectsa missing fragment then it will let the
sender know and theentire thing (all fragments) will have to be resent.

    The fragmentation is actually carried out at the IP layer not
the UDP layer.   For TCP it's 'frequently' not a problem because
the end points agree on a segment size for each packet below
the MTU thus avoiding fragmentation issues.

      I'm not sure if you guys are aware but there is not sequencing
inherit in UDP .. there's no way (unless you add one to the payload)
for the recieving system to know that packet A actually predates
packet B.   Probably not a problem since the payload will likely
have a timestamp.

James

On Monday 28 October 2002 05:21 pm, you wrote:

 >> James,
 >>
 >>   A question arose re the MIB _service_ port (as
 >> opposed to the data port).  What happens if one
 >> sends a UDP datagram that exceeds the MTU.  Is
 >> it received & reassembled correctly ?  What
 >> kind of nasty things can happen ?  Can one or
 >> more of the fragments be dropped ?  Is an error
 >> generated for that case, or is it a silent
 >> process ?  Any other upleasanteness ?
 >>
 >> Bill
-----------------------------------------------------------------------
I will also attempt to address your other points sometime
next week.

Bill

Barry Clark wrote:
>>are
>>unequivocal and unambiguous in asserting that neither TCP
>>nor UDP need concern themselves with fragmentation,
>>and that fragmentation is handled at the IP layer.
> 
> 
> Better ask James.  He has said that some switches do not do the right
> thing with fragmented UDP messages (they request resends for the fragments,
> so that it appears to work but generates excessive network trafic).  If
> our switches are of this breed, I'd still say we should stick to within 
> the MTU.
> 
> 
>>    - the Antenna ID will be a STRING
>>    - the Device ID will be a STRING
> 
> 
> For MIBs, let's store these in the slot E-Prom with the IP address.  Saves
> a lot of fooling around for the L30[12] devices.
> 
> 
>> A DDR with the alert asserted will be sent as
>>soon as the alert condition is detected, and a DDR with the alert
>>not asserted will be sent as soon as the alert condition is cleared.
>>DDRs sent between these two transitions will 1) be sent at the
>>current rate in effect for that monitor point, and 2) always
>>indicate that an alert has been asserted, which implies that
>>client software must always examine the DDR for asserted alerts.
> 
> 
> Might also be nice to reissue alerts at some slow rate (~ once a minute).
> _______________________________________________
> evla-sw-discuss mailing list
> evla-sw-discuss at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evla-sw-discuss