The DHCP server sends an ICMP echo to an IP before assigning it. "Busy" meant it got a reply to the echo, and it's marked so as to avoid an address conflict.

On 10/15/2018 10:33 AM, Ken Hohhof wrote:

Well, it started with a customer who had gotten a Mikrotik from us but it never got into the billing system which meant it was off our radar for firmware updates which meant it got hacked.  Which meant someone was using it as a proxy server and the customer complained about slow Internet.  But after replacing the Mikrotik and checking that his Internet was not “slow” anymore (these complaints are often actually WiFi issues), I noticed a ton of DHCP server entries marked “Busy”.  Not sure why the Mikrotik is doing that, I would think if the client did a release/renew, the old lease would be deleted.  Not sure what this Busy status means, except that I’m guessing that pool address won’t be available again until the least time runs out, which is bad.

Both the Hopper and the Mikrotik seem to be behaving badly, in my view.

It may be peripherally related to the situation where you replace a router, and the DHCP clients go to renew their leases and request their old address, and the new router says OK that was not the pool address I would have assigned you but it is within the range and available to what the hell, why not.  Or maybe not.  Other devices on his network, including the WiFi interface on the Hopper as well as all the Joeys, seem to have requested their old address and then kept running with no drama.

*From:* AF <[email protected]> *On Behalf Of *Nate Burke
*Sent:* Sunday, October 14, 2018 9:01 PM
*To:* AnimalFarm Microwave Users Group <[email protected]>
*Subject:* Re: [AFMUG] strange DISH Hopper2-br DHCP lease behavior

Is the customer complaining about problems, or just that you saw it and are trying to avoid issues?  Dish makes an Ethernet to MoCA Adapter, so they should be able to put that in at the router, and get rid of the Wifi to the units entirely.  I'm wondering if it's something with the Wifi Connection.

On 10/13/2018 1:21 PM, Ken Hohhof wrote:

    I noticed that a customer with DISH and a leased Mikrotik router
    from us had some strange DHCP stuff going on.  There is a lease
    for Hopper2-WiFi near the top of the pool as is normal given the
    way Mikrotik hands out addresses.  Then there are 3 different
    Joey-MoCA leases in sequence near the bottom of the pool but
    otherwise acting normally.  The weird one is Hopper2-br which
    started out near the bottom with the Joeys but every minute or two
    sends a DHCP release and then immediately a discover requesting
    the same IP address.  Apparently it’s too soon and the Mikrotik
    offers the next higher IP address.  But in another minute or two
    there is another release and request, and the IP address
    increments again.  I have the lease time set to 1 day, so at this
    rate, it is going to consume the entire address pool within a few
    hours.

    I set the DHCP lease to static and gave it 192.168.88.9 so it is
    outside the .10-.254 pool just to be safe.  The hopper is still
    doing the release/request every 1-2 minutes, but at least now it
    gets the same address again.

    Anybody else seen this?  Is this normal Hopper behavior?  Is
    something configured wrong on the router, or on the DISH
    receiver?  Maybe it doesn’t like the 1 day lease time and wants
    something shorter?

    I don’t see any wired interfaces active on the Mikrotik except the
    WAN port, so I assume the Hopper is connecting via WiFi and then
    bridging the connecting via MoCA to the Joeys.  And that
    Hopper2-WiFi is the WiFi connection, and Hopper2-br is the bridge
    to the Joeys.  Not sure why the Hopper needs 2 IP addresses, but
    apparently that is normal.  The aggressive DHCP lease behavior
    doesn’t seem right though.

    I set DHCP logging to memory, I can include logfile entries if
    that helps, but the tl;dr is a release request followed by a
    discover requesting the same IP again, so quick the timestamp
    hh:mm:ss  is the same. Then in 1-2 minutes the cycle repeats.

    I don’t know exactly why the Mikrotik was offering the next IP up
    from the one requested, I assume because it had just been released
    milliseconds before.  I don’t know if this could be criticized as
    a bug or not.  It seems like the Hopper doing a release/renew at
    1-2 minute intervals is the real problem.  If it helps, the
    Mikrotik is running 6.40.8 FW.

    (I see that 6.42.9 is out now as “long term” which I assume is the
    new name for “bugfix only”, and that 6.43.2 is out as “stable”
    which is assume is the new name for “current”.  The changelog for
    6.43 has a ton of stuff, but I have always been conservative and
    only used “bugfix only” releases.  Either 6.40.8 or 6.43.x should
    have the patch for the dreaded winbox vulnerability, and that’s my
    main concern.  With the number of changes made in 6.43, hopefully
    there will be a newer “long term” release soon.)






-- 
AF mailing list
[email protected]
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Reply via email to