[lug] OT: Wake On Lan

[D. Stimits]

Mr Viggy wrote:
> Hmm, I guess that can be confusing!  Depending on the mfgr, WOL will 
> either boot up the machine, or it will wake it from "sleep".
> 
> In either case, I found the following on an IBM web site.  I like the 
> wording, "The program sends a wake-up frame..."  The paper is from 1996, 
> though...  And it says nothing about the protocol...  However, I would 
> think that a switch/hub/router cannot, or would not, send one of these 
> packets out.  It sounds like WOL is really used to target a specific 
> machine (or a bunch, using broadcasting).  If you're having problems 
> with only one machine, I don't think the problem is WOL...
> 
> http://www.networking.ibm.com/eji/ejiwake.html

Between this and the links John Dollison has sent, I think I know how 
this is happening. The system is going through a bridge, not a router or 
gateway. The bridge transmits all broadcast packets, and many cable 
users are on this subnet. I sent a request to AT&T to ask what IP's I 
must open up to broadcast, but they never replied (I'll send another 
request and keep bugging them till they answer, or at least 
reply...telephone calls do not seem to do the trick, I will go by email 
now).

My guess is that someone in the neighborhood has WOL being transmitted, 
on the same subnet. I need to find out which addresses require broadcast 
be open to, or simply start logging broadcasts and rebooting machines to 
watch what is sent (DHCP broadcasts are apparently from normally 
non-routeable IP's). Still, I thought that this machine would ignore 
such broadcasts, I have in theory disabled WOL both in the o/s and in 
the BIOS.

D. Stimits, stimits AT attbi.com

> 
> -----------------
> How does the high-level process work?  It could work like this:
> 
> 4:47 p.m.
>     You program your PC to wake up at a specific time. Typically, you 
> schedule the wake-up request for after hours or weekends, times when 
> demands on the bandwidth are low.
> 
>     Note: The PC does not always schedule a wake-up call; the call can 
> be unsolicited if the PC and adapter are enabled.
> 
> 2:00 a.m.
>     The program sends a wake-up frame (also called a packet) over the 
> Ethernet or Token-Ring network to the Wake-on-LAN-enabled adapter 
> installed in your Wake-on-LAN-enabled PC. (A frame is a LAN transmission 
> unit that includes control and checking characters, data and delimiters.
> 
> 2:01 a.m.
>     The adapter intercepts the wake-up frame and signals your PC to 
> power up.
> 
> 2:02 a.m.
>     Your powered-down (cold) PC powers up and goes through its normal 
> startup sequence.
> 
> 2:05 a.m.
>     Working with NetFinity or other system management software, your PC 
> initiates the work that you've scheduled for it.
> 
> 4:21 a.m.
>     Your PC completes the work and waits for a period of inactivity.
> 
> 4:36 a.m.
> Your PC goes into hibernation--a sleep mode (5%-20% awake)--or, with the 
> right software, back to sleep completely by shutting down. Zzzzzzz.
> 
> Smart PC. Smart adapter. Smart you.
> ---------------------------------------------
> 
> D. Stimits wrote:
> 
>> Mr Viggy wrote:
>>
>>> Is it actually rebooting, or comming out of "sleep" mode?  My 
>>> impression was that WOL would bring a machine out of sleep, but not 
>>> reboot the system (unless you told it to).
>>
>>
>>
>> Sleep mode is not used (at least I don't think it is), it is actually 
>> starting from turned off. Perhaps I'll have to look closer and figure 
>> out if the owner has done something to cause sleep mode instead of off 
>> without my knowing. I was under the impression that WOL would work on 
>> "off" power machines if the motherboard/power supply combination 
>> supported a minimal electrical power to the network card when "off". I 
>> know that the NIC's have power on all but one local machine when 
>> turned off, because the network switch panel lights show when a NIC is 
>> active, and it is active on all machines here when they are "off" 
>> (except for the very old one that truly powers off when the off button 
>> is hit...in that case the switch indicates the link is gone). This is 
>> where I wonder if all of the manufacturers do WOL the same way, and 
>> exactly what the requirements are to kill it off and disable it; 
>> having a port I could firewall would be so darn simple if it would do 
>> the job, but I bet it is not that simple.
>>
>>>
>>> You should prolly upgrade to WinME.  98 has, well, issues.  I'm 
>>> running 98 on my "soft router" machine (I have an internal DSL card), 
>>> and about 
>>
>>
>>
>> Eeeek! [sorry :P ]
>>
>>> once a month or so, it stops all Internet activity.  I have no idea 
>>> why, it just does.  Once I reboot the system, all is fine again.  Oh, 
>>> and this system is shut off every evening (i.e. it's not running all 
>>> the time).
>>
>>
>>
>> If it were my machine I'd make it Win 2K, but it is not mine. There is 
>> not much hope of it being "upgraded" (or at least changed to a lesser 
>> evil).
>>
>>>
>>> Viggy
>>>
>>> D. Stimits wrote:
>>>
>>>> I am wondering if anyone here is familiar with how Wake On Lan (WOL) 
>>>> is communicated? In this case, it is the Linksys LNE100TX, ver. 5.1, 
>>>> but perhaps there are common ways to implement this.
>>>>
>>>> For one, is there a specific port used, or maybe a broadcast 
>>>> message? Also, does this WOL feature detect when an ethernet cable 
>>>> has been connected (I know the local switch/hub detects this), and 
>>>> run WOL just by detecting a new physical connection?
>>>>
>>>> My motive for asking this is that one machine in the newly wired 
>>>> house sometimes spontaneously turns on. WOL has been turned off in 
>>>> the BIOS as far as I can tell (the card is not integrated, so the 
>>>> BIOS may not have complete control). The system (use the word 
>>>> loosely) is Win98 on fairly modern hardware. I am trying to figure 
>>>> out why it can sometimes reboot. Originally it was found to do this 
>>>> when turning off power on the bridge and switch (testing UPS 
>>>> software), but on rare occasions it does this at other times. My 
>>>> hope is that there is something I can use the Linux filtering bridge 
>>>> for to at least remove all Internet influences, and narrow the 
>>>> debugging down to the LAN or machine itself.
>>>>
>>>> D. Stimits, stimits AT attbi.com