New
- Per-storage job limits (#154). Each local storage location and remote SSH host has a Max Concurrent Jobs field in its edit form, and an S3 offsite destination has one on its plugin configuration. Set it to 1 on a spinning disk and its jobs run one after another while other storage keeps going. Every job on a repository stored there counts: backups, restores, prunes, checks, catalog rebuilds and offsite syncs, plus the destination an offsite sync or restore copies to or from. Off by default, and never higher than the global Max Concurrent Jobs setting, so raise that first if you want more. The API accepts
max_concurrenton storage location and remote SSH create and update. Thanks to @sainf for the request.
Docker
- Ordered shutdown and service supervision (#440).
tiniis now PID 1 and the container waits on Apache, sshd, cron and MariaDB.docker stopshuts them down in order, Apache first, then MariaDB with a full flush, then ClickHouse, instead of letting Docker kill the databases when the grace period ran out. If one of those services exits on its own, the container stops with a non-zero status so your restart policy brings it back whole, rather than staying up while backups fail. There's a HEALTHCHECK for Apache, sshd and MariaDB, and the shipped compose file setsstop_grace_period: 90s; withdocker run, pass--stop-timeout 90. Thanks to @rienafairefr for the report.
No agent changes in this release.