88 Commits

Author	SHA1	Message	Date
Gan, Jimmy	6bc81f409c	ci: fix workflow stability by using standard actions	2026-04-04 23:18:24 +08:00
Gan, Jimmy	939a99ac22	fix: use full git clone in test workflow to ensure SHA is available	2026-04-04 19:54:30 +08:00
Gan, Jimmy	8aa48d26b7	fix: force remove existing container before deploying	2026-04-04 18:25:11 +08:00
Gan, Jimmy	c296a8e50a	fix: use host network for dashboard deployment in CI ... Avoid network conflicts by using host network instead of bridge networks. The runner is already in gitea_gitea network causing endpoint conflicts.	2026-04-04 18:07:12 +08:00
Gan, Jimmy	0dd836131e	fix: use docker compose down before deploying dashboard ... Use compose down --remove-orphans to properly clean up before deployment.	2026-04-04 18:00:26 +08:00
Gan, Jimmy	9fd16a42fe	fix: create required Docker networks before deploying dashboard	2026-04-04 17:55:21 +08:00
Gan, Jimmy	7af1e3cfc4	fix: create nas-dashboard directory before copying compose file	2026-04-04 17:51:07 +08:00
Gan, Jimmy	71e1991420	fix: set DOCKER_API_VERSION=1.43 for all docker commands ... The runner's Docker client version 1.53 is too new for the NAS daemon (1.43). Export DOCKER_API_VERSION=1.43 in all steps that use docker commands.	2026-04-04 17:26:53 +08:00
Gan, Jimmy	7dd2d6df9f	fix: use manual checkout instead of GitHub Actions ... Gitea Actions runner cannot fetch actions from GitHub (connection reset). Revert to manual git clone for checkout and direct tool installation.	2026-04-04 17:03:22 +08:00
Gan, Jimmy	c3ab14fa30	perf: split claude-dev into base and runtime images for faster builds ... - Create Dockerfile.base with all tools and dependencies (built once) - Slim down Dockerfile to just copy scripts from base image - CI checks for base image existence, builds if missing - Runtime builds drop from ~17min to ~10sec (only copies scripts) - Base image only needs rebuild when tools/versions change	2026-04-04 16:24:05 +08:00
Gan, Jimmy	55d3348085	fix: simplify CI workflows to run directly on runner without nested containers ... - Remove container blocks causing Docker-in-Docker permission issues - Use actions/checkout@v4 instead of manual git cloning - Use actions/setup-python@v5 and actions/setup-node@v4 for clean environments - Remove nsenter workarounds and mount complexity - Fix volume paths to use /volume1/docker/* directly - Add /volume1/docker/claude-dev to runner valid_volumes config This eliminates 20+ commits of permission/namespace hacks by letting workflows run directly in the runner's environment which already has Docker access and proper volume mounts configured.	2026-04-04 16:00:49 +08:00
Gan, Jimmy	87af015ef1	fix: use virtual environment for Python dependencies in CI	2026-04-04 15:36:09 +08:00
Gan, Jimmy	e60c5d534e	fix: use --break-system-packages for pip install in CI	2026-04-04 15:17:22 +08:00
Gan, Jimmy	0c0f3bb6d8	fix: manually clone repo from Gitea in test workflow	2026-04-04 15:15:06 +08:00
Gan, Jimmy	a7fa54476d	debug: check workspace contents	2026-04-04 15:03:40 +08:00
Gan, Jimmy	1c2ac48efd	fix: run tests without custom containers to access repo files	2026-04-04 14:33:22 +08:00
Gan, Jimmy	8629f594a3	debug: check workspace contents in test workflow	2026-04-04 14:11:56 +08:00
Gan, Jimmy	366d535f02	fix: simplify test workflow to avoid GitHub Actions dependencies	2026-04-04 13:55:14 +08:00
Gan, Jimmy	4769297ab9	fix: use host path when executing deploy script via nsenter	2026-04-04 12:44:25 +08:00
Gan, Jimmy	d588a48529	fix: write deploy script to host filesystem and execute via nsenter	2026-04-04 11:37:47 +08:00
Gan, Jimmy	8e27885ca1	fix: add mount namespace (-m) to nsenter for filesystem access	2026-04-04 11:32:37 +08:00
Gan, Jimmy	78aff7523e	fix: use full docker path in nsenter commands	2026-04-04 10:42:37 +08:00
Gan, Jimmy	f21f129a48	fix: add --pid=host to workflow container for nsenter access	2026-04-04 10:36:33 +08:00
Gan, Jimmy	3191381bbf	fix: use nsenter with privileged runner for deploy	2026-04-04 10:18:18 +08:00
Gan, Jimmy	0dbf647bb7	fix: use docker run with host network for deploy to avoid permissions issue	2026-04-04 10:11:41 +08:00
Gan, Jimmy	89d0b8d341	fix: use nsenter to run deploy in host network namespace	2026-04-04 10:02:43 +08:00
Gan, Jimmy	a401ba677c	fix: run workflow directly on host to avoid Docker-in-Docker network issues	2026-04-04 09:57:16 +08:00
Gan, Jimmy	9daf636fa0	docs: document Docker-in-Docker limitation preventing auto-deploy	2026-04-04 09:47:20 +08:00
Gan, Jimmy	b432017c84	fix: use docker rm -f pattern to avoid network reconnection issues	2026-04-04 09:41:13 +08:00
Gan, Jimmy	b594b8b37b	test: verify auto-deploy works after reverting name field	2026-04-04 09:17:21 +08:00
Gan, Jimmy	80c964db3f	fix: revert to original working deploy approach ... The 'name: nas-dashboard-dev' field added in commit a88f5b9 broke auto-deployment by causing Docker Compose network connection errors. Reverting to the original working approach: - Remove 'name:' field from docker-compose.dev.yml - Use simple 'docker compose up -d --pull never' This worked for months before the recent changes.	2026-04-04 08:35:04 +08:00
Gan, Jimmy	2a8af8feea	fix: document manual deployment requirement ... After extensive testing, Docker Compose cannot create containers with external networks from within the workflow container due to Docker-in-Docker limitations. The workflow now builds the image successfully and provides clear instructions for manual deployment. CI builds the image ✅ Manual deployment required for network-connected containers	2026-04-04 02:39:50 +08:00
Gan, Jimmy	391b7fcb54	fix: execute deploy script from mounted NAS volume ... The deploy script runs with the NAS host's Docker context, avoiding Docker-in-Docker network issues. The script is mounted at /nas-dashboard/deploy-dev.sh in the workflow container.	2026-04-04 02:26:07 +08:00
Gan, Jimmy	8514d5ea4b	fix: just restart container instead of recreating ... Docker restart will use the new image that was just built. Avoids all the network reconnection issues from recreating the container. Simple and proven to work.	2026-04-04 02:18:26 +08:00
Gan, Jimmy	431e2c0a5f	fix: execute docker commands on NAS host via SSH ... The Docker-in-Docker limitation prevents creating containers with external networks from within the workflow container. Execute the deploy commands directly on the NAS host via SSH to avoid this issue. This is similar to how manual execution works - running docker compose directly on the host where all networks are accessible.	2026-04-04 02:13:10 +08:00
Gan, Jimmy	2f65691e15	fix: use docker rm -f pattern from working claude-dev workflow ... Replicate the proven approach from deploy-claude-dev-dev.yml: - Use 'docker rm -f' to completely remove the old container - Let compose create a fresh container from scratch - Avoids the 'stop then recreate' flow that fails with network errors This works because Docker Compose handles initial network connection during 'up' differently than reconnection during recreate. The rm -f approach succeeds even from the workflow container context.	2026-04-04 02:05:09 +08:00
Gan, Jimmy	2a8e5d4952	fix: stop then up instead of force-recreate ... force-recreate still tries to disconnect/reconnect networks which fails. Use stop then up - compose will detect the image changed and recreate the container automatically without network issues.	2026-04-04 01:52:58 +08:00
Gan, Jimmy	e396f77299	fix: use --force-recreate to update container in-place ... This avoids disconnecting from networks which seems to cause issues when run from the workflow container. Force recreate updates the container without removing it first.	2026-04-04 01:36:18 +08:00
Gan, Jimmy	f27d414182	fix: cd to compose directory before running docker compose ... The workflow needs to be in the same directory as the compose file for docker compose to work properly. Manual execution worked, so replicate that approach.	2026-04-04 01:26:56 +08:00
Gan, Jimmy	f19b70d179	fix: use docker stop/rm instead of compose down for deploy ... docker compose down may have issues with external networks or project names. Use direct docker stop/rm commands instead.	2026-04-04 01:15:24 +08:00
Gan, Jimmy	e0cc38ecb5	fix: properly recreate container on deploy ... Use 'docker compose down' then 'up' instead of --no-recreate to ensure the new image is used and avoid network connection errors.	2026-04-04 00:45:55 +08:00
Gan, Jimmy	2c7787dced	revert: disable BuildKit due to slow performance on NAS ... BuildKit operations (load dockerfile, load context, FROM) are taking 20-80s each on the NAS due to filesystem/resource constraints. Legacy builder is faster. Keeping pip mirror optimization which provides the real speed improvement.	2026-04-04 00:27:39 +08:00
Gan, Jimmy	ee72a891d5	perf: optimize build speed with pip mirror and BuildKit	2026-04-04 00:19:43 +08:00
Gan, Jimmy	7a50c070e1	fix: remove duplicate docker.sock mount from deploy workflows	2026-04-03 23:49:45 +08:00
Gan, Jimmy	5e9fa2ba56	fix: add Docker socket mount to CI workflows ... - Add /var/run/docker.sock mount to deploy-dev and deploy workflows - Fixes docker build hanging indefinitely in CI - Docker commands need socket access to communicate with daemon	2026-04-03 23:14:54 +08:00
Gan, Jimmy	af5b4d8411	fix: use --no-recreate to avoid network issues, fallback to create if needed	2026-04-02 21:12:53 +08:00
Gan, Jimmy	f275e52843	fix: stop container before compose up to avoid network recreation issues	2026-04-02 10:29:31 +08:00
Gan, Jimmy	21d0cd9ff6	fix: simplify deploy to match production pattern - use idempotent docker compose up	2026-04-02 10:01:25 +08:00
Gan, Jimmy	e152de6fc8	fix: add --remove-orphans and remove force-recreate flag	2026-04-02 09:54:19 +08:00
Gan, Jimmy	b425613941	fix: use docker-compose down to clean up networks properly	2026-04-02 09:37:47 +08:00