Files
spacedrive/.github/workflows/server.yml
T
Jamie Pineandcapy-ai-staging[bot] 5bc474e722 CLI and daemon housekeeping from the live Mac runs (#3111)
* Default the replication block in core.status so a newer CLI reads an older daemon

A 10dbc18 CLI asking a pre-#3109 daemon for status over --device failed
with 'missing field replication'. ReplicationStatus now derives Default
and the field carries serde(default); a unit test deserializes the
pre-#3109 payload shape. ts-client types regenerated.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Expose replication.paused through sd config get, set and show

The pause switch was only readable through sd status; config get reported
an unknown key. get reads it from the daemon config, set calls the
existing mounts.replication_set_paused action, show lists it, and the
help text names every key get and set accept.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Answer sync traffic for an absent library once per connection, with an Error reply

A member that deleted its copy of a library stays in the other members'
device table, so every catch-up they ran logged an ERROR on the receiver
and timed out on the sender. The multiplexer now logs one WARN per
connection (keyed on the peer's connected_at) and debug after that,
answers requests with the existing SyncMessage::Error variant, and the
backfill request helpers turn that reply into a failure with the peer's
reason so the catch-up loop backs off instead of marking the peer caught
up. No wire change.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Read paired and connected counts in core.status from the network registry

core.status hardcoded both to zero, so sd status said Paired 0 while
sd network status said Paired 1. Both now read the device registry,
addresses come from the endpoint, and the rows are labelled
'Paired devices' and 'Connected devices'.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Resumable, logged sd update with sd update status

Downloads land in <data>/updates/<asset>.<build>.part and resume with a
Range request on the next attempt; a transfer that moves no bytes for
60 s is abandoned for that run with the part kept; the sha256 is checked
on the complete part before either binary is replaced. Every attempt
appends a line with its result and reason to logs/update.log and writes
update-state.json; sd update status prints the running build, the
sd-version.txt marker, the last attempt, any part waiting to resume and
the timer state. After an install the build commit (or release tag) is
written to sd-version.txt beside the binaries, the same marker the local
rebuild script writes. The launchd agent already had RunAtLoad; the docs
now say so.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Ask launchd for 65536 file descriptors in the daemon agent

The plist set 8192 and the core warned about it at every boot.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* List jobs without loading each job's persisted state blob

jobs.list read full job rows, including the serialized job state; an
indexer over millions of files keeps its seen-path set there, so a
hundred rows took 45 s over a remote link and not much less locally.
The listing and the running-job fallback now select only the columns
the row needs through a partial model. The payload is unchanged.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Drop maximize-build-space from the Blacksmith release and server jobs

The action leaves a Blacksmith runner with about 100 MB of root disk.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Register the housekeeping row in PROJECT_STATUS.md

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Make main the branch of record: nightly and core tests trigger on main

main was fast-forwarded to sources at 10dbc18. nightly.yml and
core_tests.yml now trigger on main, with sources kept in the lists for
one release so a stray push there still builds; the nightly docs and
the PROJECT_STATUS.md register row name main.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

* Bound every updater HTTP read with the idle timeout

The checksum and release fetches awaited their bodies with no timeout,
so a stalled proxy could hang a timer run before anything was logged.

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>

---------

Co-authored-by: capy-ai-staging[bot] <305915985+capy-ai-staging[bot]@users.noreply.github.com>
2026-10-05 19:54:56 +00:00

121 lines
3.7 KiB
YAML

name: Server release
on:
push:
tags:
- "v*"
workflow_dispatch:
jobs:
build-server:
name: Build a docker image for spacedrive server
runs-on: blacksmith-4vcpu-ubuntu-2404
defaults:
run:
shell: bash
steps:
- name: Checkout repository
uses: actions/checkout@v4
- name: Set up Homebrew
id: set-up-homebrew
uses: Homebrew/actions/setup-homebrew@master
- name: Update Podman & crun
run: |
sudo apt-get remove crun podman
brew install crun podman
- name: Update buildah
shell: bash
run: |
set -euxo pipefail
wget -O- 'https://github.com/HeavenVolkoff/buildah-static/releases/latest/download/buildah-amd64.tar.gz' \
| sudo tar -xzf- -C /usr/local/bin
- name: Install netavark
shell: bash
run: |
set -euxo pipefail
sudo mkdir -p /usr/local/lib/podman
sudo wget -O- 'https://github.com/containers/netavark/releases/latest/download/netavark.gz' \
| gunzip | sudo dd status=none of=/usr/local/lib/podman/netavark
sudo chmod +x /usr/local/lib/podman/netavark
- name: Install passt
shell: bash
working-directory: /tmp
run: |
set -euxo pipefail
deb="$(
curl -SsL https://passt.top/builds/latest/x86_64 \
| grep -oP 'passt[^\.<>'\''"]+\.deb' | sort -u | head -n1
)"
curl -SsJLO "https://passt.top/builds/latest/x86_64/${deb}"
sudo dpkg -i "${deb}"
- name: Basic sanity check
run: |
crun --version
podman version
buildah --version
- name: Determine image name & tag
id: image_info
shell: bash
run: |
set -euxo pipefail
if [[ "$GITHUB_REF" == refs/tags/* ]]; then
IMAGE_TAG="${GITHUB_REF##*/}"
# Only tag as "latest" for stable releases (no alpha/beta/rc suffix)
if [[ "$IMAGE_TAG" =~ ^v[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
EXTRA_TAG="latest"
else
EXTRA_TAG=""
fi
else
IMAGE_TAG="$(git rev-parse --short "$GITHUB_SHA")"
EXTRA_TAG="staging"
fi
IMAGE_TAG="${IMAGE_TAG,,}"
IMAGE_NAME="${GITHUB_REPOSITORY,,}/server"
echo "Building ${IMAGE_NAME}:${IMAGE_TAG}"
echo "tag=${IMAGE_TAG}" >> "$GITHUB_OUTPUT"
echo "extra_tag=${EXTRA_TAG}" >> "$GITHUB_OUTPUT"
echo "name=${IMAGE_NAME}" >> "$GITHUB_OUTPUT"
echo "repo=${GITHUB_REPOSITORY}" >> "$GITHUB_OUTPUT"
echo "repo_ref=${GITHUB_HEAD_REF:-${GITHUB_REF#refs/heads/}}" >> "$GITHUB_OUTPUT"
- name: Set up QEMU
uses: docker/setup-qemu-action@v3
with:
platforms: 'arm64'
- name: Build image
id: build-image
uses: redhat-actions/buildah-build@v2
with:
tags: ${{ steps.image_info.outputs.extra_tag && format('{0} {1}', steps.image_info.outputs.tag, steps.image_info.outputs.extra_tag) || steps.image_info.outputs.tag }}
archs: amd64, arm64
image: ${{ steps.image_info.outputs.name }}
layers: 'false'
context: .
build-args: |
REPO=${{ steps.image_info.outputs.repo }}
REPO_REF=${{ steps.image_info.outputs.repo_ref }}
containerfiles: |
./apps/server/Dockerfile
- name: Push image to ghcr.io
uses: redhat-actions/push-to-registry@v2
with:
tags: ${{ steps.build-image.outputs.tags }}
image: ${{ steps.build-image.outputs.image }}
registry: ghcr.io
username: ${{ github.actor }}
password: ${{ secrets.GITHUB_TOKEN }}