Skip to content

ansible/dnsmasq: work around a busybox bug

Martin Roukala requested to merge dnsmasq_aaaa_dns_fail into master

When booting duts in vivian, one may see the following issue:

[4.03]: Synchronize the clock
+ ntp_set ci-gateway
+ peer_addr=ci-gateway
+ status=DONE
+ time timeout 5 ntpd -dnq -p ci-gateway
+ Command terminated by signal 15
real	0m 5.00s
user	0m 0.00s
sys	0m 0.00s
+ status=FAILED
+ log 'Getting the time from the NTP server ci-gateway: FAILED'
[9.05]: Getting the time from the NTP server ci-gateway: FAILED

When checking the dnsmasq logs, we can see that the query worked for ipv4, but failed for ipv6:

Mar 03 11:00:48 vivian dnsmasq[615827]: query[A] ci-gateway from 10.42.0.3
Mar 03 11:00:48 vivian dnsmasq[615827]: config ci-gateway is 10.42.0.1
Mar 03 11:00:48 vivian dnsmasq[615827]: query[AAAA] ci-gateway from 10.42.0.3
Mar 03 11:00:48 vivian dnsmasq[615827]: config error is REFUSED (EDE: not ready)

The only solution I found to this problem was to tell dnsmasq to always return the invalid ipv6 address for AAAA requests... and it seems everyone is happy.

Merge request reports