Ddns works fine but system (dnsmasq?) uses some old IPs


I put my fritzbox 7360 a little bit away, it only works as a VDSL modem now.
I took my TP-LINK AC1750 (archer C7), put LEDE Reboot 17.01.4 r3560-79f57e422d / LuCI lede-17.01 branch (git-17.290.79498-d3f0685) (is this correct actual version?).
It has 4 ports LAN, 2 WLAN and a WAN connector. The fritzbox is connected as a VDSL modem to, the WAN and LEDE does pppoe.

The whole setup works fine for all LAN clients and WLAN clients, perfect.

However, system uses weird IP for a ddns address:

my ddns for zappa.mig19.de works fine:

090837       : Detect registered/public IP
 090837       : #> /usr/bin/nslookup zappa.mig19.de  >/var/run/ddns/myddns_ipv4.dat 2>/var/run/ddns/myddns_ipv4.err
 090837       : Registered IP '' detected
 090837  WARN : Updating IP at DDNS provider failed - starting retry 2/0
 090837       : Detect local IP on 'network'
 090837       : Local IP '' detected on network 'wan'
 090837       : Update needed - L: '' <> R: ''
 090837       : #> /bin/uclient-fetch -q -O /var/run/ddns/myddns_ipv4.dat -Y off 'https://CUSTOM_SECRET_URL' 2>/var/run/ddns/myddns_ipv4.err
 090838       : DDNS Provider answered:
 090838  info : Update successful - IP '' send
 090838  info : Forced update successful - IP: '' send
 090838       : Waiting 600 seconds (Check Interval)

The new IP is used and updated everywhere on the internet, currently for example ping.eu correctly says the following:

--- PING zappa.mig19.de ( 56(84) bytes of data. ---
64 bytes from icmp_seq=1 ttl=58 time=20.4 ms
64 bytes from icmp_seq=2 ttl=58 time=20.3 ms
64 bytes from icmp_seq=3 ttl=58 time=20.3 ms
64 bytes from icmp_seq=4 ttl=58 time=20.5 ms

Double check ifconfig, if is correct:

pppoe-wan Link encap:Point-to-Point Protocol
          inet addr:  P-t-P:  Mask:


Bot local system uses some weird old address:

root@archer-c7:~# ping zappa.mig19.de
PING zappa.mig19.de ( 56 data bytes
64 bytes from seq=0 ttl=63 time=33.186 ms

If I stop dnsmasq system uses actual correct address but connected clients have no DNS anymore at all.

I appreciate your help very much, is there something wrong with dnsmasq?

Kind Regards

Okay, meanwhile I found out one part of a solution is to use the option dns_server in the ddns config part. When not the own dnsmasq is queried with the old IP but for example (which has actual entry from successful previous run) the part "[...]Updating IP at DDNS provider failed[...] vanishes and it recognizes a previous successful run.

However, root cause is that dnsmasq is keeping an old entry for the dynamic host for ever. How do I update my own entry or make this automatically updated every couple of minutes or so? So actually the caching of dnsmasq is ultra aggressive.

In other words: DDNS runs fine now, system (dnsmasq) keeps old address which seems for ever.

Solution: server=/mig19.de/ into /etc/dnsmasq.conf.
I do not know if there is a better kind of automatic solution.