Description of problem: As described at https://gitlab.labs.nic.cz/labs/knot/issues/493 knotd when HUP-reloading tries to load the original symlink's destination, not the default file. There is a patch to be included in the next release. AFAICT, there is no schedule yet on when that might be. Can you include the patch directly? Version-Release number of selected component (if applicable): > knotd --version Knot DNS, version 1.6.8 > rpm -q knot knot-1.6.8-1.el7.x86_64 How reproducible: always Steps to Reproduce: sh-4.2$ # knot.conf is a valid config file sh-4.2$ cd /tmp/ sh-4.2$ mkdir orig sh-4.2$ mv knot.conf orig/ sh-4.2$ ln -s orig/knot.conf . sh-4.2$ vi knot.conf sh-4.2$ knot^C sh-4.2$ mkdir new^C sh-4.2$ mv orig new sh-4.2$ ln -fs new/knot.conf . sh-4.2$ kill -HUP 113 With the following knotd output (logging to stderr). sh-4.2$ knotd -c /tmp/knot.conf Warning: log severity 'all' is no longer supported. Using 'info' instead. 2016-12-19T13:15:33 info: Knot DNS 1.6.8 starting 2016-12-19T13:15:33 info: binding to interface '0.0.0.0@53536' 2016-12-19T13:15:33 info: binding to interface '::@53536' 2016-12-19T13:15:33 info: configured 2 interfaces and 0 zones 2016-12-19T13:15:33 info: loading zones 2016-12-19T13:15:33 warning: no zones loaded 2016-12-19T13:15:33 info: starting server 2016-12-19T13:15:33 info: server started in the foreground, PID 113 2016-12-19T13:15:33 info: binding remote control interface to '/tmp/knot/knot.sock' 2016-12-19T13:16:14 info: reloading configuration 2016-12-19T13:16:14 error: configuration file '/tmp/orig/knot.conf' not found
*** This bug has been marked as a duplicate of bug 1406194 ***