Five scopes, one winner
Set backup.retention at any of shc’s five config scopes; the narrowest scope that defines it wins, and the running daemon hot-reloads the change without a restart.
-
global cluster 30d shipped default
-
tenant acme - unset
-
environment prod - unset
-
stack blog - unset
-
deployment blog@prod - unset
effective
backup.retention
30d
← won at global (widest, nothing narrower is set)
shcd● runningpid 4211 · node-a
uptime2d 11:04:00← never resets
livebackup.retention =30d
config_loader · DB watch ~5s
$ shc config ls --prefix backup.
Demonstrates shc config set backup.retention 7d --scope tenant -t acme · shc config get backup.retention -t acme -s blog · shc config unset: the 5-level DB cascade (global ⊃ tenant ⊃ environment ⊃ stack ⊃ deployment, narrowest wins), hot-reloaded by config_loader’s ~5s DB watch firing V200100CFGCHG. HUD key C.
simulated. The real thing: one config table in the cluster DB that every shcd watches and applies live, no restart.