User-agent: Mozilla Thunderbird 0.8 (Windows/20040913)
The behaviour is unpredictable, but it is not related to the clients.
Other systems have no problems keeping track of the same services. It
must be our use of the mDNS APIs or perhaps our platform bindings. Is
there any documentation on how to correctly implement the StartBrowse
callback?
Stuart Cheshire wrote:
Hi,
I'm tracking down a problem where services disappear and re-appear for
unknown reasons every few hours or so. I noticed in our debug log that
the callback from mDNS_StartBrowse() is invoked with AddRecord == false
while the answer->rroriginalttl > 0.
In which case should the service be removed? When AddRecord == false or
when the TTL reaches 0? The comments in the header file are a little
ambiguous about this...
Have fun,
Arthur van Hoff
answer->rroriginalttl is the original value that was in the record when
it was received on the wire. (Information that's actually less useful
than one might imagine.)
The best way to dignose this problem is probably with an mDNSNetMonitor
trace.
If the service is *really* not answering queries, then after four failed
queries, it will be removed from the cache. An mDNSNetMonitor trace would
show that.
When you say, "disappear and re-appear", how long is the interval between
disappearing and re-appearing? A second? A minute? (Running "dns-sd -B"
in a terminal window will give you a time-stamped log of when things
appear and disappear.)
Stuart Cheshire <email@hidden>
* Wizard Without Portfolio, Apple Computer, Inc.
* www.stuartcheshire.org
--
Arthur van Hoff, email@hidden
Tivo Inc. 2160 Gold Street Alviso, CA 95002
408-519-9207 (tivo), 650-283-0842 (cell), 408-519-5339 (fax)
