dan at milkcarton.com
Wed Apr 16 18:28:35 EDT 2003
Chris, you are partialy right on the packet to the mac address issue. I
WOL enabled nic will wake up the host when it gets a *magic packet*
addressed to its mac address. Do a google on magic packet if you
want/need more info on it.
Chris Palmer wrote:
>Christian Hoenig wrote
>>>>You then first try to contact them, and if that fails, you try to
>>>>wake them up.
>>>I don't think you have that kind of control. When you attempt
>>>to make contact, WoL (if enabled) would cause the target machine
>>>to wake because of the incoming packets. Either you send a packet
>>>to that host or you don't.
>>I'm not quite sure, if I missunderstand you, or you missunderstand WOL.
>>If you enable wol on a network adapter, it only wakes up, if you send the
>>'magic' paket to this adapter. It does not wake up, if you send any paket to
>maybe I misunderstand WOL, then. I was under the impression that
>it would automatically wake the host if it saw a packet that was
>addressed to its mac address.
>>>Another issue that may need to be looked at is that the frontend
>>>doesn't currently attempt to re-establish a connection to the
>>>backend, after communication is lost, which is annoying in
>>>general, but could make it impossible to implement this idea.
>>I don't see the issue with that thing? :-)
>>What do you think is the problem?
>I don't know if there has been work on it this week (still catching
>up on the forums), but if the frontend and backend stopped responding
>to each other (different hosts, or same), then the frontend never
>re-establishes its connection with the backend. The only way to
>make things work again is to restart the frontend.
>So... it all depends on your shutdown/sleep/bootup/wakeup flow
>on how this will work. If you're doing all this wakeup stuff
>while attempting your initial "connect" from the frontend to
>the backend, then you will not run into this problem. If the
>backend sleeps on you when your frontend is still up (but idle)
>then "re-connection" needs to be negotiated.
>Maybe I should drop out of this thread, since I don't know
>what the runtime flow looks like for your implementation of
>suspend/wakeup, and I'm just confusing myself... :)
>heh... if you get it working, and it applies to my setup,
>I'll be glad to test it... ;)
>mythtv-dev mailing list
>mythtv-dev at snowman.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mythtv-dev