Resolver failures after stale-answer enabled
Today I got bug reports from our ISP customers. Their browser often
fail at page loading with DNS_PROBE_FINISHED_NXDOMAIN. After few
seconds (possibly after browser DNS re-query) page will load correctly. In
Bind9 log I see many of messages like:
Oct 27 11:34:13 srv-snv-production named[576109]: configuration.ls.apple.com resolver failure, stale answer unavailable
Oct 27 11:34:13 srv-snv-production named[576109]: client @0x7fc71806cd58 10.202.42.196#58876 (configuration.ls.apple.com): view clients: query failed (SERVFAIL) for configuration.ls.apple.com/IN/TYPE65 at query.c:5832
Oct 27 11:34:13 srv-snv-production named[576109]: configuration.ls.apple.com resolver failure, stale answer unavailable
Oct 27 11:34:13 srv-snv-production named[576109]: client @0x7fc7180715a8 10.202.42.196#49219 (configuration.ls.apple.com): view clients: query failed (SERVFAIL) for configuration.ls.apple.com/IN/A
at query.c:5832After I turned off stale-answer-enable, problem looks to be resolved. I'm
attaching huge debug log of above failures. The problematic query starts at 27-Oct-2021 11:34:13.858
Linux srv-le-production 5.10.0-9-amd64 #1 SMP Debian 5.10.70-1 (2021-09-30) x86_64 GNU/Linux
bind9/stable,now 1:9.16.15-1 amd64