none
WoL using Unicast RRS feed

  • Question

  •  

    Hi!

     

    Is there anyone who is using SCCM with the Wake-on-lan feature? I'm in a medium/large sized company with around 10,000 clients and want to use the built in wake-on-lan feature (with unicast). My problem when testing is that clients only wake up a few hours after shut down.

     

    Example:

    Computer A has been off for 1 hour and B has been off 24 hours. When distributing software to both A and B (and checking the "Enable Wake on lan") - only A wakes up. 

     

    I think (after a bit of reading) this has to do with the ARP cache on the routers between the Site server and the client which loses the clients MAC-adress after a while. Read somewhere that you could increase the ARP cache size in Windows? Would doing so only on the Site server solve the problem or do you have to do something with the routers as well? Or is this a completely different problem? Network protocols and routers are not my cup of tea...

     

    Thanks

    Orange, Sweden

    Friday, January 11, 2008 4:12 PM

Answers

  • Yes, from what little hardware experience I have, it is in minutes (30 or less). So you'd have to configure the routers to have a longer ARP cache life.

     

    We do expect that people will use this feature. However, it does not have all the features that some 3rd party products might have, we know that. It is a version 1 product feature.

     

    Friday, January 11, 2008 11:40 PM

All replies

  • The target client information has to be in the ARP cache of the site server or router in order for WOL to work. You'd increase the ARP cache life in the router, not in Windows.

     

    Friday, January 11, 2008 6:44 PM
  • Thank you for your time and swift response Wally!

     

    OK, let's see if I understand this correctly (remember I usually don't have to worry about deep technical network issues so bear with me please); The ARP cache in Windows is "locked" so you get no help from there. Then your only option is to increase the cache life in the routers so they keep "remember" the adresses of the clients.

     

    But isn't this ARP "Time to live" value usually counted in minutes and hours? So even if you greatly increase that value - you can never wake up clients which have been shut down for, say, two weeks? That maybe isn't necessary but I'd love to have at least 12 hours.

     

    In short; is this WoL feature is doable in a company of this size? It would also be really intresting to know a little more exactly how Microsoft intended this feature to work in reality.

     

    Thank you!

    Orange

     

    Friday, January 11, 2008 9:30 PM
  • Check to see if the router has WOL disabled, if you don't control every router from point to point there is a good chance it is killed in trasit.

    To overcome this problem we created a program to package the WOL packets and then open with at the destination on a machine and send it out.  1E and other vendors have applications that do the same thing.
    Friday, January 11, 2008 10:40 PM
    Moderator
  • Yes, from what little hardware experience I have, it is in minutes (30 or less). So you'd have to configure the routers to have a longer ARP cache life.

     

    We do expect that people will use this feature. However, it does not have all the features that some 3rd party products might have, we know that. It is a version 1 product feature.

     

    Friday, January 11, 2008 11:40 PM
  • I have also had same problems when using unicast WOL.

     

    Here are a couple of comments:

    - Cisco's default ARP cache life is 4 hours (at least IOS 12.0). With IOS command 'arp timeout 0', you could define ARP cache entries as static, so they would never be removed from cache.

    http://www.cisco.com/en/US/docs/ios/12_0/np1/command/reference/1ripadr.html#wp1017517

    I have not idea how many ARP cache entries could be in a routers ARP cache. Probably this is a BAD idea...

    - if the computer is within same subnet as site server, you could add the target computer's MAC address as a static entry to site server's ARP cache with a command ARP -s. This is only useful when testing unicast based WOL within one subnet.

     

    I've had difficult time finding the exact knowledge how unicast WOL should work and which packets should wake the computer up. Finally, I found an interesting document: Network Device Class Power Management Reference Specification, v. 2.0

    (http://www.microsoft.com/whdc/resources/respec/specs/pmref/PMnetwork.mspx)

     

    It defines how network adapters should handle wake up events. Here is a quotation from document's Appendix A:

     

    "The following describes the minimum set of patterns for a machine with one IP address per adapter and a computer name.  It is the absolute minimum set of patterns that would be useful to wake up on. If the NIC can support more than this set, they will be used for additional IP addresses, multicast addresses, or NetBIOS names.

     

    Notes:

    ·        'computername' below is "WAKER"

    ·        Station address is 08003e304770

    ·        NetBIOS scope is null

    ·        Media is 802.3 without SNAP.  For a different framing (SNAP, FDDI, ARCNET etc.) the offsets would be different.

     

    ARP to machine address 157.55.199.72

    This pattern will wake a machine up whenever another machine needs to obtain its MAC address.  This generally precedes any IP transaction.

     

    Offset:                           Hex Pattern:

    =====================================================================

    12        0806                                              ; Protocol type (0806 = ARP)

    21        01                                                  ; ARP Opcode (01 = Request)

    38        9d37c748          ; IP Address requested (157.55.199.72)

     

    Directed IP packet (note that this excludes any other directed packets to our MAC address)

    This pattern will wake the machine up if any other machine sends information directly to it (Normal IP transactions).

     

    Offset:                           Hex Pattern:

    =====================================================================

    0          08003e304770 ; Destination MAC Address (Station Address)

    12        0800                                              ; Protocol type (0800 = IP)

    30        9d37c748          ; IP Address (157.55.199.72)

     

    NBT Name Query/Registration for computername <00>, <03>, <20>

    This pattern will wake a machine up whenever another machine needs to obtain its IP address in order to initiate an IP Transaction (using NetBIOS)

     

    Offset:                           Hex Pattern:

    =====================================================================

    12        0800                                                                         ; Protocol type (0800 = IP)

    23        11                                                                             ; Protocol (11= UDP)

    34        00890089                                     ; Port Number (NETBIOIS Name Service)

    45        10                                                                             ; NetBIOS Flags (10 = Query OR Registration)

    54                                  20                                                            ; Name scope – NULL (limited to 32 bytes)

    55                                  46 48 45 42 45 4C 45 46                     ; Computerame – coded in half-ASCII (‘WAKER’)

    56                        46 43 43 41 43 41 43 41 ; **Note that this field is always 30 bytes.  The final character

    57                        43 41 43 41 43 41 43 41                             ; is not included because it is not unique (may be 00, 03 or 20)

    79        43 41 43 41 43 41

    "

     

    So this tells that ARP queries should wake the computer up. This would enable to wake up computers in remote subnets even if their MAC address is not in router's ARP cache. However, computers are not waken up when they receive an ARP query. Now I'm confused if this could be considered as a bug with network adapter's driver implementation...

    And I'm not sure if this document is the current specification that network adapters should follow.

    Thursday, February 28, 2008 2:10 PM
    Moderator
  • As long as the target system's MAC address is in the site server's ARP cache, or the router's, then we should be able to wake it up. If not, then it is a limitation of the current release of our WOL solution.

    Thursday, February 28, 2008 4:34 PM
  • Yes, the main question for getting this to work "in reality" is how many hours you can stretch the ARP cache life without causing other problems on the network. And what type of errors a too long ARP cache life would generate. I understand it's different for every type and size of the network - but if someone has some sort of a general guideline or a hint what to expect it would be nice.

     

    Because the WoL function itself really works excellent in all my tests (if the computers haven't been down for more than 4 hours of course).

     

    /Orange

     

     

    Sunday, March 2, 2008 10:36 PM
  • That would be a question for some network guys, not SMS guys :-)

     

    I am not aware of any issues with a long ARP cache. I know customers have set it to multiple hours (usually default to minutes) without issues. I don't have any other thoughts to offer. Other than requiring more memory for a larger ARP cache, I can't think of any issues with this.

     

    Monday, March 3, 2008 12:09 AM
  •    Prerequisites for Wake On LAN

    http://technet.microsoft.com/en-us/library/bb680822.aspx

    "You can configure child primary sites for Wake On LAN, but not secondary sites. Clients in secondary sites will be sent wake-up packets from their assigned primary site."

     

     

    1. There is a reason for not sending WOL packets from Secondary Sites?

    2. What is a "child primary site"?

    3. Does "Child primary site" need aditional license?

    4. How can I add\configure a "child primary site" to my Primary Site? (I need these "child" for a remote branch which is in another subnet).

     

    Thank you!

    Friday, April 11, 2008 3:39 PM
  •  

    A child Primary is another primary site under an existing primary site.  It is a primary since it has SQL running while a Secondary doesn't.

     

     

    Yes you can place this primary in the other subnet and use it for WOL

     

    Friday, April 11, 2008 4:03 PM
    Moderator
  • Hi Mathew,

    When you say this:

    To overcome this problem we created a program to package the WOL packets and then open with at the destination on a machine and send it out.  1E and other vendors have applications that do the same thing.

    What do you  mean and ow do you do it?

    -h-

    -If Life gives you lemons, make lemonade
    Tuesday, June 2, 2009 5:32 AM
  • Yes, I would also like to know the details of this.

    To overcome this problem we created a program to package the WOL packets and then open with at the destination on a machine and send it out.  1E and other vendors have applications that do the same thing.

    Thanks!!!

    Friday, April 30, 2010 4:14 PM