Repository navigation
feat(server)!: configure the server with server.toml and run workers on Podman - #1470
Merged
Merged
Conversation
Launchers name their location and are resolved once at start-up. BREAKING CHANGE: --launchers, --oidc-config and --trusted-proxy-config are replaced by sqb server --config server.toml with [[launchers.docker]], [[launchers.local]], [oidc] and [trusted_proxy] sections. The old flags fail with a pointer to --config.
…ob resources Docker workers keep the location container's dropped capabilities and security options, take [launchers.docker.run] memory, cpus, pids and cgroup_parent, and a job's resources, with max_memory as the ceiling. Only positive limits are sent, so Podman workers are not started with a zero process limit. The reference image can install a released SQLBuild.
sqb location health prints a target's state and exits 1 when no instance has checked in for 90 seconds or its last load failed. Docker workers join the location's named network when Podman reports it as bridge. A path-filtered and nightly Verify job runs the packaged Compose lifecycle on rootful Podman. sqb dev and sqb location get documentation pages.
kvlonge
enabled auto-merge
November 14, 2026 03:46
… job GitHub's Ubuntu runner keeps /run/podman private to root, so changing the socket's group alone left the runner user unable to connect.
Podman has no BuildKit API, so buildx on the runner built in its own BuildKit container and the images never reached Podman's store.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
Running the server in containers, and on Podman in particular, meant several config files, a pile of flags, and workers that ignored container limits. Container health checks also needed a way to tell when a code location was down or falling back to an older load.
Changes
sqb serverandsqb scheduler --config server.tomlread[server],[retention],[oidc],[trusted_proxy]and[launchers.*]from one file, and a flag overrides the file. Launchers name their location, which is resolved when the server starts. BREAKING CHANGE:--launchers,--oidc-configand--trusted-proxy-configare replaced by--config.resources(memory, cpus, pids), then the launcher's[launchers.docker.run]table, then the location container.max_memorycaps them; without it, a job can't go above the location container's limits. Workers keep the location's dropped capabilities and security options.sqb location healthprints a target's state and exits 1 when the location is down or falling back, for use in container health checks.GET /api/locations/{id}/healthtakes an optionaltarget.server.toml, worker limits, Podman and the newsqb devandsqb locationpages. DuckShop gains a container resources edge case.Verification
make check-cipassed on the final head.sqb location healthcase that fails without the per-target read. Orchestration native tests passed 133/133.podman composestack ran a worker.