Compare commits

...
5 Commits
Author SHA1 Message Date
admin d88eac4c65 ci(connectivity_tester): use inputs.tag for dispatch 2026-09-16 23:12:13 +02:00
admin 4c9bc90dcb ci(connectivity_tester): add Gitea Actions release workflow 2026-09-15 22:21:43 +02:00
hermes f071c7fb43 fix(hermes_infra/gitea): add dedicated act_runner cache-server (#34)
## Problem

The built-in `act_runner` cache server binds to an interface unreachable from per-job container networks (`container.network: ""` is the default). `ACTIONS_CACHE_URL` ends up pointing at an unreachable IP, and every CI job burns ~9 minutes on:

```
::warning::Failed to restore: getCacheEntry failed: connect ETIMEDOUT 192.168.112.21:37051
::warning::Failed to save:    reserveCache  failed: connect ETIMEDOUT 192.168.112.21:37051
```

Measured on `admin/brencher#15` (https://gitea.rudolf.keenetic.link/admin/brencher/actions/runs/16):

| Job     | wall-clock | of which dead air on cache |
|---------|------------|----------------------------|
| 3.12.3  | 12m 40s    | ~9m (271s × 2 timeouts)    |
| 3.13    | 11m 13s    | ~9m                        |
| 3.14    | 11m 17s    | ~9m                        |

So even though `python-package.yml` enables `cache: 'pip'`, the cache backend is unreachable, jobs hit `pip install` cold every time, and CI ends up ~4x slower than necessary.

## Fix

Run a dedicated `act_runner cache-server` container on the same overlay network as the runner, and point the runner at it via `external_server`. Shared secret comes from a Docker secret mounted into both containers and read via `ACT_RUNNER_EXTERNAL_SECRET_FILE`.

## Changes

- `poc/gitea/stack-compose.yml`:
  - New `cache_server` service (`command: ["cache-server"]`) on the `gitea` overlay with a `cache_data` named volume.
  - `runner` and `cache_server` both mount the external Docker secret `gitea_cache_secret` at `/run/secrets/gitea_cache_secret` and set `ACT_RUNNER_EXTERNAL_SECRET_FILE` accordingly.
  - Top-level `secrets:` block declares `gitea_cache_secret` as `external: true`, with the create command commented inline next to it.
- `poc/gitea/config.yaml`: point `cache.{host,port,external_server}` at `gitea_cache_server:37051`; `external_secret` left empty because the secret is loaded from the file.

## Verification

- Both YAMLs parse (PyYAML).
- Service name `gitea_cache_server` resolves on the `gitea` overlay via Swarm service-name DNS.
- `docker secret inspect gitea_cache_secret` works after the one-time create step.

## Deploy notes

1. On the swarm manager, create the secret once:
   ```
   echo -n "$(openssl rand -hex 32)" | docker secret create gitea_cache_secret -
   ```
2. `docker stack deploy -c stack-compose.yml gitea`.
3. Re-run a CI job and confirm cache `restore`/`save` log lines no longer ETIMEDOUT.

## Related

- Related: admin/brencher#15 — `python-package.yml` enables `cache: 'pip'` and was timing out for the reason this PR fixes.
- Refs: gitea/actions/runs/16 jobs #248/249/250.

## Supersedes

- Supersedes: admin/ansible#33 — first attempt was based on a stale `origin/master` and accidentally included `hermes_squid/` changes from the actual upstream merge. The current branch is rebased on `admin/master` (`c3b7128`) and only touches `poc/gitea/`.Reviewed-on: http://192.168.1.96/admin/ansible/pulls/34
Reviewed-by: admin <rudolfss13@gmail.com>
Co-authored-by: hermes <rudolfss13+hermes@gmail.com>
2026-09-04 08:11:41 +02:00
c3b7128006 hermes_squid: open docker_socket_proxy to full read-only scope (#4)
Flip CONTAINERS/IMAGES/NETWORKS/VOLUMES/SERVICES/TASKS/NODES from 0 to 1 so hermes-agent can inspect running stacks (read container logs, enumerate swarm services) without going through each node's shell.

POST stays 0 — no container creation, no image builds, no exec into running containers. Anything that mutates state remains blocked at the proxy level.

---------

Co-authored-by: admin <rudolfss13@gmail.com>
Co-authored-by: hermes <hermes@rudolf.keenetic.link>
Reviewed-on: http://192.168.1.96/admin/ansible/pulls/4
Co-authored-by: hermes <rudolfss13+hermes@gmail.com>
2026-08-30 00:09:56 +02:00
hermesandhermes 0c95d8e1ee feat(hermes_squid): enable local STT via faster-whisper (runtime install) (#21)
## What

Wires up local STT (faster-whisper) into the hermes-gateway container so voice messages on Telegram transcribe without depending on external services.

Closes #19. Supersedes the closed #20 (which used a custom Dockerfile-hermes; this version uses a bind-mounted entrypoint script instead since pypi.org is now whitelisted).

## Changes

1. **`hermes_squid/squid.conf`** — expands the egress whitelist:
   - new `stt_pkg_domains` for `pypi.org`, `files.pythonhosted.org`, `huggingface.co`, `github.com`, `models.dev`, `registry.npmjs.org`, `cdn-lfs.huggingface.co` — needed for runtime `pip install faster-whisper` and HF model fetch.
   - new `telegram_ips` covering AS62041 (91.108.4.0/22, 91.108.8.0/22, 91.108.12.0/22, 91.108.16.0/22, 91.108.20.0/22, 91.108.56.0/22, 149.154.160.0/20, 185.76.151.0/24) as a backstop for the `dstdomain` ACL. squid's `dstdomain` does not match HTTPS CONNECT requests without `ssl_bump`, so the recent log shows 24 of 9083 `api.telegram.org` requests getting 403 in a 5-minute window.
   - renamed `hermes_network` -> `localnet` (squid default name, no behavior change).

2. **`hermes_squid/hermes-gateway-entrypoint.sh`** (new) — bind-mounted into the container at `/opt/data/hermes-gateway-entrypoint.sh` and set as the service entrypoint. The script:
   - is idempotent: on first start, runs `apt-get install -y python3-pip ffmpeg` and `pip3 install --break-system-packages faster-whisper`. Subsequent restarts skip install if `faster_whisper` is already importable.
   - exports `STT_PROVIDER=local`, `STT_LOCAL_MODEL=base`, `STT_LOCAL_DEVICE=cpu`, `STT_LOCAL_COMPUTE_TYPE=int8`.
   - `exec`s the original gateway command (`gateway run`).

3. **`hermes_squid/stack-compose.yml`** — `hermes-gateway` gets:
   - `entrypoint: ["/opt/data/hermes-gateway-entrypoint.sh"]`
   - `command: ["gateway", "run"]`
   - bind-mount `./hermes-gateway-entrypoint.sh:/opt/data/hermes-gateway-entrypoint.sh:ro`
   - new env: `STT_PROVIDER`, `STT_LOCAL_MODEL`, `STT_LOCAL_DEVICE`, `STT_LOCAL_COMPUTE_TYPE`

No other service in the compose file is touched.

## Why runtime install, not a custom image

- `pypi.org` is now whitelisted, so `pip3 install` works at runtime through the same squid.
- Avoids maintaining a custom hermes-agent image and a local registry tag.
- The upstream `nousresearch/hermes-agent:v2026.8.3` image is kept as-is.

## Deploy notes

- brencher apply will trigger `docker service update` for hermes-gateway; the new entrypoint runs the apt+pip install on first start (~1-2 min).
- First start also downloads the `base` faster-whisper model (~150 MB) from `huggingface.co` — covered by `stt_pkg_domains`.
- squid config is reread on `docker service update` for the squid service; new ACLs apply without a full stack restart.
- Subsequent restarts of hermes-gateway skip the install path entirely (idempotent guard).

## Verified locally

- `git diff --stat upstream/master` shows only the three files above (+25 entrypoint, +14 squid, +8 compose).
- The new worktree is `/opt/data/work/thread-114-Alex-InfraAlexBot/ansible` (per-session working dir).

## Risks

- `apt-get update` inside the container needs `deb.debian.org` reachable — already in `allowed_domains`.
- `STT_LOCAL_MODEL=base` is ~150 MB and runs on CPU. If latency on Raspberry-class nodes is bad, switch to `tiny`.
- `--break-system-packages` is required for PEP 668 (Debian trixie + Python 3.11+). If the upstream image switches to PEP 668 strictly, this flag is the right escape hatch.

## Related

- #19 — investigation that found missing faster-whisper + squid deny pattern
- #20 — superseded (used Dockerfile-hermes instead of runtime install)

---------

Co-authored-by: hermes <hermes@rudolf.keenetic.link>
Reviewed-on: http://192.168.1.96/admin/ansible/pulls/21
Reviewed-by: admin <rudolfss13@gmail.com>
Co-authored-by: hermes <rudolfss13+hermes@gmail.com>
2026-08-29 22:53:46 +02:00
8 changed files with 216 additions and 32 deletions
+57
View File
@@ -0,0 +1,57 @@
name: release
on:
push:
tags:
- 'v*'
workflow_dispatch:
inputs:
tag:
description: Tag to build (e.g. v0.1.0-test1)
required: true
default: v0.1.0-test1
jobs:
build:
runs-on: ubuntu-latest
steps:
- name: checkout
uses: actions/checkout@v4
- name: log into Gitea Packages
uses: docker/login-action@v3
with:
registry: gitea.rudolf.keenetic.link
username: ${{ secrets.REGISTRY_USERNAME }}
password: ${{ secrets.REGISTRY_TOKEN }}
- name: resolve version
id: version
run: |
if [ -n "${{ github.event.inputs.tag }}" ]; then
echo "VERSION=${{ github.event.inputs.tag }}" >> "$GITHUB_OUTPUT"
else
echo "VERSION=${GITHUB_REF_NAME#refs/tags/}" >> "$GITHUB_OUTPUT"
fi
- name: build and push
uses: docker/build-push-action@v6
with:
context: connectivity_tester
file: connectivity_tester/Dockerfile
push: true
tags: |
gitea.rudolf.keenetic.link/hermes/connectivity_tester:${{ steps.version.outputs.VERSION }}
gitea.rudolf.keenetic.link/hermes/connectivity_tester:latest
- name: render stack-compose
run: |
mkdir -p out
TAG=${{ steps.version.outputs.VERSION }}
sed "s|\${version}|$TAG|g" connectivity_tester/stack-compose.yml > "out/docker-compose-$TAG.yml"
- name: upload rendered stack-compose
uses: actions/upload-artifact@v4
with:
name: stack-compose
path: out/
+5 -1
View File
@@ -21,6 +21,9 @@ server {
listen 8080;
server_name _;
resolver 127.0.0.11 192.168.1.1 valid=30s ipv6=off;
resolver_timeout 5s;
# Tiny access log; default nginx format
access_log /var/log/nginx/brencher-state-access.log;
error_log /var/log/nginx/brencher-state-error.log warn;
@@ -37,7 +40,8 @@ server {
# The only allowed path. Exact match: query strings or trailing
# anything else fall through to the default 403.
location = /state {
proxy_pass https://brencher.rudolf.keenetic.link/state;
set $brencher_upstream https://brencher.rudolf.keenetic.link;
proxy_pass $brencher_upstream/state;
proxy_http_version 1.1;
# Upstream is HTTPS; the sidecar trusts the LE-issued cert via
+9 -5
View File
@@ -1,16 +1,20 @@
acl hermes_network src 10.0.0.0/8
acl hermes_network src 172.16.0.0/12
acl hermes_network src 192.168.0.0/16
acl localnet src 10.0.0.0/8
acl localnet src 172.16.0.0/12
acl localnet src 192.168.0.0/16
acl allowed_domains dstdomain .minimax.io .minimax.chat .minimax.com .githubusercontent.com deb.debian.org gitea.rudolf.keenetic.link
acl telegram_domains dstdomain .telegram.org .t.me
acl stt_pkg_domains dstdomain .pypi.org .files.pythonhosted.org .huggingface.co .github.com .models.dev .registry.npmjs.org .cdn-lfs.huggingface.co .hf.co .xethub.hf.co .cas-server.xethub.hf.co
acl telegram_ips dst 91.108.4.0/22 91.108.8.0/22 91.108.12.0/22 91.108.16.0/22 91.108.20.0/22 91.108.56.0/22 149.154.160.0/20 185.76.151.0/24
acl CONNECT method CONNECT
acl SSL_ports port 443
acl Safe_ports port 80 443
http_access deny !Safe_ports
http_access deny CONNECT !SSL_ports
http_access allow hermes_network allowed_domains
http_access allow hermes_network telegram_domains
http_access allow localnet allowed_domains
http_access allow localnet telegram_domains
http_access allow localnet stt_pkg_domains
http_access allow localnet telegram_ips
http_access deny all
http_port 3128
+7 -7
View File
@@ -169,13 +169,13 @@ services:
# Anything else (POST, CONTAINERS, IMAGES, SERVICES, ...) must be explicitly
# toggled to 1 by a consumer that has a reason to need it. NEVER widen the
# scope from a generic change like this one.
CONTAINERS: 0
IMAGES: 0
NETWORKS: 0
VOLUMES: 0
SERVICES: 0
TASKS: 0
NODES: 0
CONTAINERS: 1
IMAGES: 1
NETWORKS: 1
VOLUMES: 1
SERVICES: 1
TASKS: 1
NODES: 1
POST: 0
ports:
- target: 2375
+39
View File
@@ -0,0 +1,39 @@
#!/usr/bin/env bash
# Custom entrypoint for the act_runner cache-server.
#
# The upstream image's /usr/local/bin/run.sh hardcodes `act_runner daemon`
# and ignores argv. So we replace the entrypoint here.
#
# cache-server also requires cache.external_secret to be set in config.yaml
# as a literal value (not a file path). We keep the value in a Docker
# secret and substitute it into a runtime copy of config.yaml before
# exec'ing act_runner.
set -eu
SECRET_FILE="${ACT_RUNNER_EXTERNAL_SECRET_FILE:-/run/secrets/gitea_cache_secret}"
CONFIG_FILE="${CONFIG_FILE:-/cfg/config.yaml}"
if [[ ! -s "$SECRET_FILE" ]]; then
echo "cache-server-run: secret file $SECRET_FILE is missing or empty" >&2
exit 1
fi
if [[ ! -f "$CONFIG_FILE" ]]; then
echo "cache-server-run: config file $CONFIG_FILE not found" >&2
exit 1
fi
SECRET_VALUE="$(cat "$SECRET_FILE")"
# Render a copy of config.yaml with cache.external_secret replaced by the
# secret's literal value. /cfg/config.yaml is a read-only Swarm config mount,
# so we cannot edit it in place. Use a plain-file replacement — the
# upstream config.yaml has the form `external_secret: ""` (or any literal)
# on a single line, so a simple sed is safe.
RUNTIME_CONFIG="/tmp/config.runtime.yaml"
cp "$CONFIG_FILE" "$RUNTIME_CONFIG"
# Replace the value after `external_secret:` on a single line.
# Use a delimiter unlikely to appear in the secret.
sed -i "s|^ external_secret:.*$| external_secret: \"${SECRET_VALUE}\"|" "$RUNTIME_CONFIG"
exec /usr/local/bin/act_runner cache-server --config "$RUNTIME_CONFIG"
+8 -18
View File
@@ -70,26 +70,16 @@ runner:
- "ubuntu-22.04:docker://docker.gitea.com/runner-images:ubuntu-22.04"
cache:
# Enable cache server to use actions/cache.
enabled: true
# The directory to store the cache data.
# If it's empty, the cache data will be stored in $HOME/.cache/actcache.
dir: ""
# The host of the cache server.
# It's not for the address to listen, but the address to connect from job containers.
# So 0.0.0.0 is a bad choice, leave it empty to detect automatically.
host: ""
# The port of the cache server.
# 0 means to use a random available port.
port: 0
# The external cache server URL. Valid only when enable is true.
# If it's specified, act_runner will use this URL as the ACTIONS_CACHE_URL rather than start a server by itself.
# The URL should generally end with "/".
# Requires external_secret below to be set to the same value on both this runner and the cache-server.
external_server: ""
# Shared secret between this runner and the external `act_runner cache-server`. Required when external_server
# (or `act_runner cache-server`) is in use: the runner pre-registers each job's ACTIONS_RUNTIME_TOKEN with the
# cache-server, and the cache-server enforces bearer auth + per-repo cache isolation.
# Point at the dedicated gitea_cache_server container on the `gitea` overlay.
# The runner's built-in cache server binds to an interface unreachable from
# per-job container networks, causing ~9 min/job wasted on
# "reserveCache failed: connect ETIMEDOUT" (see gitea/actions/runs/16 job #248).
host: "gitea_cache_server"
port: 37051
external_server: "http://gitea_cache_server:37051/"
# Read from the gitea_cache_secret Docker secret (see stack-compose.yml).
external_secret: ""
container:
+36
View File
@@ -0,0 +1,36 @@
#!/usr/bin/env bash
# Custom entrypoint for the gitea_runner service.
#
# Same trick as cache-server-run.sh: act_runner requires cache.external_secret
# to be a literal string in config.yaml, not a file path. We keep the value in
# a Docker secret and substitute it into a runtime copy of config.yaml, then
# exec the image's own /usr/local/bin/run.sh (which handles registration +
# daemon).
set -eu
SECRET_FILE="${ACT_RUNNER_EXTERNAL_SECRET_FILE:-/run/secrets/gitea_cache_secret}"
CONFIG_FILE="${CONFIG_FILE:-/cfg/config.yaml}"
if [[ ! -s "$SECRET_FILE" ]]; then
echo "runner-run: secret file $SECRET_FILE is missing or empty" >&2
exit 1
fi
if [[ ! -f "$CONFIG_FILE" ]]; then
echo "runner-run: config file $CONFIG_FILE not found" >&2
exit 1
fi
SECRET_VALUE="$(cat "$SECRET_FILE")"
# Render a runtime copy with the secret substituted in.
RUNTIME_CONFIG="/tmp/config.runtime.yaml"
cp "$CONFIG_FILE" "$RUNTIME_CONFIG"
sed -i "s|^ external_secret:.*$| external_secret: \"${SECRET_VALUE}\"|" "$RUNTIME_CONFIG"
# The upstream run.sh reads $CONFIG_FILE from the environment to pass
# --config to act_runner. Override it so registration + daemon both pick up
# the runtime copy with the substituted secret.
export CONFIG_FILE="$RUNTIME_CONFIG"
exec /usr/local/bin/run.sh
+55 -1
View File
@@ -1,4 +1,8 @@
# Gitea Actions stack. Uses a dedicated act_runner cache-server so that
# job-container networks can reach the cache backend via Swarm service DNS
# (gitea_cache_server:37051) instead of the unreachable IP the built-in
# cache server binds to per-job.
services:
server:
image: docker.gitea.com/gitea:1.27.0
@@ -19,6 +23,7 @@ services:
runner:
image: docker.io/gitea/act_runner:nightly
entrypoint: ["/usr/local/bin/runner-run.sh"]
environment:
CONFIG_FILE: /cfg/config.yaml
GITEA_INSTANCE_URL: "https://gitea.rudolf.keenetic.link"
@@ -26,14 +31,51 @@ services:
GITEA_RUNNER_NAME: "runner1"
GITEA_RUNNER_LABELS: "runner_label1"
# GITEA_RUNNER_EPHEMERAL: "1"
ACT_RUNNER_EXTERNAL_SECRET_FILE: /run/secrets/gitea_cache_secret
networks:
- gitea
# - nginx_ingress
configs:
- source: runner_config
target: /cfg/config.yaml
mode: 0444
- source: runner_run
target: /usr/local/bin/runner-run.sh
mode: 0755
volumes:
- /var/run/docker.sock:/var/run/docker.sock
secrets:
- gitea_cache_secret
deploy:
restart_policy:
condition: on-failure
delay: 5s
max_attempts: 3
cache_server:
# The image's /usr/local/bin/run.sh hardcodes `act_runner daemon` and
# ignores argv, so passing `cache-server` via `command` doesn't help —
# the daemon path is always taken. We replace the entrypoint by
# mounting our own script via a Swarm config (file mount), then have it
# substitute the Docker secret literal into a runtime copy of
# config.yaml's cache.external_secret and exec act_runner.
image: docker.io/gitea/act_runner:nightly
entrypoint: ["/usr/local/bin/cache-server-run.sh"]
environment:
ACT_RUNNER_EXTERNAL_SECRET_FILE: /run/secrets/gitea_cache_secret
networks:
- gitea
volumes:
- cache_data:/root/.cache/actcache
secrets:
- gitea_cache_secret
configs:
- source: runner_config
target: /cfg/config.yaml
mode: 0444
- source: cache_server_run
target: /usr/local/bin/cache-server-run.sh
mode: 0755
networks:
nginx_ingress:
@@ -44,6 +86,17 @@ networks:
configs:
runner_config:
file: ./config.yaml
cache_server_run:
file: ./cache-server-run.sh
runner_run:
file: ./runner-run.sh
secrets:
gitea_cache_secret:
# Create once before first deploy (or to rotate):
# echo -n "$SECRET" | docker secret create gitea_cache_secret -
# Then `docker stack deploy -c stack-compose.yml gitea`.
external: true
volumes:
data:
@@ -52,4 +105,5 @@ volumes:
type: nfs
o: addr=192.168.1.96
device: :/root/volumes/gitea
# device: :/media/DATA/runtime/gitea
# device: :/media/DATA/runtime/gitea
cache_data: