Re: Can an app see the same UDP packet multiple times?

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

Could that be another instance of famous last words :)...
8-bit packet ID is dangerously narrow. I'd play safe and
use a 32-bit value instead.

BTW, speaking from experience, sometimes 32-bit packet IDs
may not be enough either...

--
=====================================
Alexander Nickolov
Microsoft MVP [VC], MCSD
email: agnickolov@xxxxxxxx
MVP VC FAQ: http://www.mvps.org/vcfaq
=====================================

"Dean Roddey" <droddey@xxxxxxxxxxxxxxxx> wrote in message
news:uPv4r2WIGHA.3000@xxxxxxxxxxxxxxxxxxxxxxx
>I went ahead and bit the bullet and implemented a rejection scheme. Each
>outgoing event gets a unique id, based on an MD5 hash id generated for each
>application when it loads (behind their backs of course, down in the event
>send/receive subsystem), and then the low byte of that hash is set to an
>incrementing sequence number for each event that goes out. It'll wrap at
>255, but that's way more than long enough to handle duplicate rejection. It
>would be a truely pathological application that sent more than 255 events
>in less than a few minutes. The receiving thread is keeping the last 128
>ids it's seen, once in a hash table for fast lookup and another copy in a
>deque so it can quickly toss the oldest id when a new one comes in. I think
>that 128 will be more than enough, but I can increase it if needed. If it's
>seen the id, it just doesn't drop the new event into the processing queue.
>
> -------------------------------------
> Dean Roddey
> Chairman/CTO, Charmed Quark Systems
> www.charmedquark.com
>
> "m" <m@xxx> wrote in message
> news:%23JkJClSIGHA.3120@xxxxxxxxxxxxxxxxxxxxxxx
>> Yes, this is likly the cause - but don't rely on it.
>>
>> As has been mentioned aready, any arbitrary packet may arrive zero or
>> more times.
>>
>> "Dean Roddey" <droddey@xxxxxxxxxxxxxxxx> wrote in message
>> news:Om8oeZRIGHA.3936@xxxxxxxxxxxxxxxxxxxxxxx
>>> I'm assuming that it's happening because the user has multiple network
>>> adaptors, all on the local subnet, one of them wireless, and that I'm
>>> listening on 0.0.0.0 so as to avoid having to bind to a particular
>>> adapter. The more of them he enables, the more 'echos' he sees. So I
>>> think I'm just seeing them coming in on multiple adapters. The rest of
>>> us, who have single adapter systems, aren't seeing this problem.
>>>
>>> -------------------------------------
>>> Dean Roddey
>>> Chairman/CTO, Charmed Quark Systems
>>> www.charmedquark.com
>
>


.

Relevant Pages