Took 5 days for the issue to surface again. It seems that any change to the EPMP AP that causes a radio reset will clear the issue. But disassociating or rebooting the SM will not. The last time I changed the 'Reliable multicast' setting and it started working. This time I changed the frame size from 2.5ms to 5ms and it started working. But both of those changes caused the RF to reset, so I don't think that it's the setting so much as something in the AP network/rf stack getting reset. I'll open a ticket with Cambium, but as infrequent as it is, I'm not sure what they'll be able to find.

On 3/22/2022 11:17 AM, Nate Burke wrote:
It has not happened again since Saturday, but that's the only AP out of the hundred deployed that I set that 'reliable multicast' setting on. I think that it did cause an RF Drop when that setting was changed, so maybe just kicking the radio from the AP would be enough? Every other site with the issue had multiple sectors, so kicking the SM From one AP would have it attach to another AP at the site and it would start working. However rebooting the SM and it attaching back to the same AP does not fix it. Need to wait another couple days for it to happen again to do some more testing.Â

On 3/22/2022 7:59 AM, dmmoff...@gmail.com wrote:

Interesting. I think that initial offer is a broadcast packet…..maybe that “reliable multicast� setting affects broadcast as well?

Â

*From:*AF <af-boun...@af.afmug.com> *On Behalf Of *Nate Burke
*Sent:* Saturday, March 19, 2022 7:38 AM
*To:* AnimalFarm Microwave Users Group <af@af.afmug.com>
*Subject:* Re: [AFMUG] EPMP1000 and DHCP failures

Â

I was able to get a packet capture while it while it was happening. Client had been running fine for about 3 days before it started erroring (3 hour DHCP Lease). Nothing was logged with the firewall rules on the Mikrotik doing the DHCP Server.

I have a mikrotik between the 450SM an the EPMP AP, I was able to run a packet capture from there. I ran it on the interface of the EPMP radio. It was showing the DHCP Discover being sent to the Server, and the DHCP Offer being sent back to the client, but that was it, no DHCP Request Packet coming from the EPMP Interface.Â

On the EPMP AP, I changed the 'Reliable Multicast' from Disabled to Enabled. And the client immediately got a DHCP lease after saving that (No AP Reboot). The DHCP Request Packet came back from the client as soon as the Discover/offer packets were sent. I'm not convinced that was the issue, as I don't have it enabled on ay other EPMP radio on the network. It seemed more like making a change in the AP reset something in the EPMP network stack. It's just strange that it happens so randomly.

On 3/14/2022 7:24 PM, Steve Jones wrote:

    The mikrotik that handles the dhcp relay or dhcp, log any input
    firewall rules and see if its dropping the packets

    Â

    On Mon, Mar 14, 2022, 7:03 PM Nate Burke <n...@blastcomm.com
    <mailto:n...@blastcomm.com>> wrote:

        Just had it happen on a newly installed EPMP1000<->EPMP1000
        link.  AP and SM are both 2.4 non-GPS radios.  Feed to site
        is a 450B off a450M AP.  Relay from barn to house using 2.4
        EPMP 1000 radios.Â

        Was working fine when I left,  3 hours later, DHCP lease
        timed out (Mikrotik DHCP Lease time) and would not get new
        lease.  Rebooting the 1000 Radio acting as the AP fixed
        it.  If it happens again, I'll try to get a packetcapture
        off it.

        On 3/9/2022 10:14 AM, Steve Jones wrote:

            the mikrotik is dhcp relay, BMIÂ is the dhcp server

            Â

            On Wed, Mar 9, 2022 at 10:07 AM Josh Luthman
            <j...@imaginenetworksllc.com
            <mailto:j...@imaginenetworksllc.com>> wrote:

                Oh this is on the DHCP server, sorry.

                Â

                On Wed, Mar 9, 2022 at 10:31 AM Steve Jones
                <thatoneguyst...@gmail.com> wrote:

                    we have to have it for dhcp relay to keep
                    functioning. otherwise it periodically stops
                    working from EPMP APs, I never knew why, mikrotik
                    had no answer, but it would suddenly get caught
                    up in non ACL drops add action=accept chain=input
                    comment="ALLOW DHCP UDP 67" dst-port=67
                    log-prefix=dhcp protocol=udp

                    Â

                    On Wed, Mar 9, 2022 at 8:12 AM Josh Luthman
                    <j...@imaginenetworksllc.com> wrote:

                        The input chain is to the Mikrotik itself, ie
                        the IP address that it would theoretically
                        get from the DHCP server.  I was thinking of
                        a managed Mikrotik as a demarc to the
                        customer's stuff (so forward chain).

                        Â

                        On Tue, Mar 8, 2022 at 7:57 PM Steve Jones
                        <thatoneguyst...@gmail.com> wrote:

                            I had this issue a long time ago, id like
                            to think that it was a firmware revision
                            that resolved the issue, but it was a
                            long time ago and im partially retarded.Â

                            If you have a mikrotik, add an input rule
                            allow udp 67. Just for kicks. It might be
                            this issue that i have that policy for.

                            Â

                            On Tue, Mar 8, 2022, 4:22 PM Josh Luthman
                            <j...@imaginenetworksllc.com
                            <mailto:j...@imaginenetworksllc.com>> wrote:

                                Raise a ticket with Cambium and
                                explain the situation?  If you could
                                get pcap that would show what's
                                missing. Do you have a Tik behind
                                any SM with the issue by chance?

                                Â

                                On Tue, Mar 8, 2022 at 4:05 PM Nate
                                Burke <n...@blastcomm.com> wrote:

                                    No DHCP Relay, just local DHCP
                                    Server on the mikrotik on the
                                    bridge that all the AP's are part
                                    of.

                                    No MAC limit on the SM'sÂ

                                    When it exhibits itself, a
                                    customer who has been running for
                                    weeks will timeout their lease,
                                    and the mikrotik will just go to
                                    'offered'Â Rebooting the AP
                                    always fixes it.Â

                                    On 3/8/2022 1:18 PM,
                                    dmmoff...@gmail.com wrote:

                                        I was wondering about
                                        broadcast rate limit.  That
                                        would apply to a DHCP
                                        discover, but not to a
                                        renewal.  ….but either the
                                        MAC limit or broadcast limit
                                        would clear when rebooting
                                        the SM, and he says rebooting
                                        the SM has no effect.

                                        Â

                                        Is DHCP running on the port
                                        that the AP is plugged into,
                                        or is there a DHCP relay
                                        involved?

                                        Â

                                        Â

                                        *From:* AF
                                        <af-boun...@af.afmug.com> *On
                                        Behalf Of *Josh Luthman
                                        *Sent:* Tuesday, March 08,
                                        2022 12:43 PM
                                        *To:* AnimalFarm Microwave
                                        Users Group <af@af.afmug.com>
                                        *Subject:* Re: [AFMUG]
                                        EPMP1000 and DHCP failures

                                        Â

                                        Do you have the SM limited on
                                        MACs? Look at Ethernet Port
                                        Security on config > network.

                                        Â

                                        On Tue, Mar 8, 2022 at 12:32
                                        PM Nate Burke
                                        <n...@blastcomm.com> wrote:

                                            I've experienced this
                                            issue randomly, and
                                            haven't been able to track
                                            down a cause. Wondering
                                            if anyone else has come
                                            across something similar.

                                            Mikrotik DHCP Server.Â
                                            EPMP1000 GPS AP,  Force
                                            300 SM.

                                            At a random time, one or
                                            More Force 300 SM's on
                                            the AP will lose the
                                            ability to hand out a
                                            DHCP Address to the
                                            client.  The Mikrotik just
                                            shows 'Offered'

                                            Rebooting or powercycling
the SM has no effect. If the SM Connects to a
                                            different sector, then
                                            DHCP is immediately
                                            handed out.  If the AP
                                            reboots, and the SM
                                            reconnects, then DHCP is
                                            immediately handed out.  If
                                            the SM is set for NAT
                                            mode, it can get a DHCP
                                            Address just fine, but
                                            switching back to bridge,
                                            the Customer router will
                                            not get DHCP.

                                            I've experienced this
                                            from 4.4.3 all the way up
                                            to 4.6.3.  It always
                                            seems to be an EPMP1000
                                            AP with a Foce300 SM, but
                                            does not affect every
                                            Force300 SM at the same time.

                                            At least now I know when
                                            I start having this
                                            problem to go reboot the AP.

-- AF mailing list
                                            AF@af.afmug.com
                                            
http://af.afmug.com/mailman/listinfo/af_af.afmug.com



                                    Â

-- AF mailing list
                                    AF@af.afmug.com
                                    <mailto:AF@af.afmug.com>
                                    
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

-- AF mailing list
                                AF@af.afmug.com <mailto:AF@af.afmug.com>
                                
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

-- AF mailing list
                            AF@af.afmug.com <mailto:AF@af.afmug.com>
                            http://af.afmug.com/mailman/listinfo/af_af.afmug.com

-- AF mailing list
                        AF@af.afmug.com <mailto:AF@af.afmug.com>
                        http://af.afmug.com/mailman/listinfo/af_af.afmug.com

-- AF mailing list
                    AF@af.afmug.com <mailto:AF@af.afmug.com>
                    http://af.afmug.com/mailman/listinfo/af_af.afmug.com

-- AF mailing list
                AF@af.afmug.com <mailto:AF@af.afmug.com>
                http://af.afmug.com/mailman/listinfo/af_af.afmug.com



        Â

-- AF mailing list
        AF@af.afmug.com <mailto:AF@af.afmug.com>
        http://af.afmug.com/mailman/listinfo/af_af.afmug.com



Â







-- 
AF mailing list
AF@af.afmug.com
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Reply via email to