mirror of
https://github.com/jmcorgan/fips.git
synced 2026-10-30 07:37:36 +00:00
Johnathan Corgan
c8c32cccc1
The remove and purge checks restart systemd-resolved after planting the DNS routing file, and that restart came seconds after several others: fips-dns-teardown and fips-dns-setup when the gateway block restarts the daemon, the teardown again when the remove check stops fips-dns, and postrm on the remove. By default systemd refuses more than five starts of a unit within one 10 s start-limit window, so depending on where that window opened, the purge check's restart was refused, resolved was left failed with start-limit-hit, and the resolvectl check after it waited 25 s on D-Bus activation and failed before the purge ran. The restart was unchecked, so nothing showed why. Clear the unit's failed state and start counter before restarting it, check the restart, wait for resolved to be active, and on failure record the restart's output with the unit's status and journal instead of running the resolvectl check against a resolver that is not there.