dockerize snapshot upload

.github/workflows/release-snapshot-service-image.yml

@@ -0,0 +1,62 @@
+name: Snapshot Service Image
+
+# Cancel workflow if there is a new change to the branch.
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: ${{ github.ref != 'refs/heads/main' }}
+
+on:
+  push:
+    branches: [main]
+    tags:
+      - "*"
+  merge_group:
+  pull_request:
+    branches: [main]
+
+jobs:
+  build-and-push-docker-image:
+    name: Build images and push to GHCR
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    steps:
+      - name: List cached docker images
+        run: docker image ls
+
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Login to Github Packages
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      # This step yields the following labels
+      # - date+sha, e.g. 2023-01-19-da4692d,
+      # - tag (if pushed).
+      - name: Docker Meta
+        id: meta
+        uses: docker/metadata-action@v5
+        with:
+          images: ghcr.io/chainsafe/forest-snapshot-service
+          flavor: |
+            latest=false
+          tags: |
+            type=raw,value={{date 'YYYY-MM-DD'}}-{{sha}}
+            type=ref,event=tag
+            type=edge
+
+      - name: Build image and push to GitHub Container Registry
+        uses: docker/build-push-action@v5
+        with:
+          context: ./images/snapshot-service/
+          build-contexts: |
+            common=./tf-managed/scripts/
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          push: ${{ github.ref == 'refs/heads/main' || startsWith(github.ref, 'refs/tags/v') }}
+
+      - name: List docker images
+        run: docker image ls

images/snapshot-service/Dockerfile

@@ -0,0 +1,25 @@
+# Snapshot service Dockerfile.
+# It is meant to produce a single snapshot of the given chain in the Filecoin network and upload it to S3 (preferably Cloudflare R2, it should work for other providers as well, but it wasn't tested).
+FROM docker:24
+LABEL org.opencontainers.image.description "Forest snapshot service generator and uploader for Filecoin"
+
+RUN apk add --no-cache \
+        ruby=3.2.2-r1 \
+        ruby-dev=3.2.2-r1 \
+        docker=24.0.7-r0 \
+        aws-cli=2.13.25-r0 \
+        bash=5.2.21-r0 && \
+    gem install \
+        docker-api:1.28.0 \
+        slack-ruby-client:2.2 \
+        activesupport:7.0.8
+
+COPY ./src /opt/snapshot-service
+
+# TODO: Change this once `sync-check` is fully-dockerized as well.
+# hadolint ignore=DL3022
+COPY --from=common ruby_common /opt/snapshot-service/ruby_common
+
+WORKDIR /opt/snapshot-service
+
+CMD ["bash", "run.sh"]

images/snapshot-service/README.md

@@ -0,0 +1,14 @@
+# TODO MOAR
+## Variables
+
+```bash
+R2_ACCESS_KEY=
+R2_SECRET_KEY=
+R2_ENDPOINT=
+
+SLACK_API_TOKEN=
+SLACK_CHANNEL=
+
+NETWORK_CHAIN=
+FOREST_TAG=
+```

tf-managed/modules/daily-snapshot/service/daily_snapshot.rb → images/snapshot-service/src/daily_snapshot.rb

@@ -8,24 +8,16 @@
 require 'logger'
 require 'fileutils'
 
-BASE_FOLDER = get_and_assert_env_variable 'BASE_FOLDER'
 SLACK_TOKEN = get_and_assert_env_variable 'SLACK_API_TOKEN'
-CHANNEL = get_and_assert_env_variable 'SLACK_NOTIF_CHANNEL'
-
-# Prune logs files(txt) older than 2 weeks
-def prune_logs(logs_folder = 'logs')
-  cutoff_date = Date.today - 14 # set the cutoff date to 14 days ago
-
-  Dir.glob("#{logs_folder}/*.txt").each do |file|
-    File.delete(file) if File.file?(file) && File.mtime(file).to_date < cutoff_date
-  end
-end
+CHANNEL = get_and_assert_env_variable 'SLACK_NOTIFICATION_CHANNEL'
 
 CHAIN_NAME = ARGV[0]
 raise 'No chain name supplied. Please provide chain identifier, e.g. calibnet or mainnet' if ARGV.empty?
 
 # Current datetime, to append to the log files
 DATE = Time.new.strftime '%FT%H:%M:%S'
+
+FileUtils.mkdir_p 'logs'
 LOG_EXPORT_SCRIPT_RUN = "logs/#{CHAIN_NAME}_#{DATE}_script_run.txt"
 LOG_EXPORT_DAEMON = "logs/#{CHAIN_NAME}_#{DATE}_daemon.txt"
 LOG_EXPORT_METRICS = "logs/#{CHAIN_NAME}_#{DATE}_metrics.txt"
@@ -46,7 +38,7 @@ def prune_logs(logs_folder = 'logs')
 
 upload_cmd = <<~CMD.chomp
   set -o pipefail && \
-  timeout --signal=KILL 8h ./upload_snapshot.sh #{CHAIN_NAME} #{LOG_EXPORT_DAEMON} #{LOG_EXPORT_METRICS} | \
+  timeout -s SIGKILL 8h ./upload_snapshot.sh #{CHAIN_NAME} #{LOG_EXPORT_DAEMON} #{LOG_EXPORT_METRICS} | \
   #{add_timestamps_cmd}
 CMD
 
@@ -71,6 +63,3 @@ def prune_logs(logs_folder = 'logs')
 [LOG_EXPORT_SCRIPT_RUN, LOG_EXPORT_DAEMON, LOG_EXPORT_METRICS].each do |log_file|
   logger.info "Snapshot export log:\n#{File.read(log_file)}\n\n" if File.exist?(log_file)
 end
-
-# Prune logs files(txt) in the logs directory older than 2 weeks
-prune_logs

images/snapshot-service/src/run.sh

@@ -0,0 +1,17 @@
+#!/bin/bash
+
+set -euo pipefail
+
+# Assert that all required environment variables are set
+: "${R2_ACCESS_KEY:?}"
+: "${R2_SECRET_KEY:?}"
+: "${SLACK_API_TOKEN:?}"
+: "${SLACK_NOTIFICATION_CHANNEL:?}"
+: "${NETWORK_CHAIN:?}"
+: "${FOREST_TAG:?}"
+
+aws configure set default.s3.multipart_chunksize 4GB
+aws configure set aws_access_key_id "$R2_ACCESS_KEY"
+aws configure set aws_secret_access_key "$R2_SECRET_KEY"
+
+ruby daily_snapshot.rb "$NETWORK_CHAIN"

tf-managed/modules/daily-snapshot/service/upload_snapshot.sh → images/snapshot-service/src/upload_snapshot.sh

@@ -83,7 +83,6 @@ else
   timeout 30m forest-tool snapshot validate --check-links 0 --check-network "$CHAIN_NAME" --check-stateroots 5 forest_db/forest_snapshot_*.forest.car.zst
 fi
 
-
 # Kill the metrics writer process
 kill %1
 
@@ -95,24 +94,35 @@ CONTAINER_NAME="forest-snapshot-upload-node-$CHAIN_NAME"
 docker stop "$CONTAINER_NAME" || true
 docker rm --force "$CONTAINER_NAME"
 
-CHAIN_DB_DIR="$BASE_FOLDER/forest_db/$CHAIN_NAME"
-CHAIN_LOGS_DIR="$BASE_FOLDER/logs"
+CHAIN_DB_DIR="/opt/forest_db/$CHAIN_NAME"
+CHAIN_LOGS_DIR="/opt/logs/$CHAIN_NAME"
+mkdir -p "$CHAIN_DB_DIR"
+mkdir -p "$CHAIN_LOGS_DIR"
 
-# Delete any existing snapshot files. It may be that the previous run failed
-# before deleting those.
-rm "$CHAIN_DB_DIR/forest_snapshot_$CHAIN_NAME"*
+# Cleanup volumes from the previous if any.
+DB_VOLUME="${CHAIN_NAME}_db"
+LOG_VOLUME="${CHAIN_NAME}_logs"
+docker volume rm "${DB_VOLUME}" || true
+docker volume rm "${LOG_VOLUME}" || true
 
-# Run forest and generate a snapshot in forest_db/
+# Run forest and generate a snapshot in the `DB_VOLUME` volume.
 docker run \
   --name "$CONTAINER_NAME" \
   --rm \
   --user root \
-  -v "$CHAIN_DB_DIR:/home/forest/forest_db":z \
-  -v "$CHAIN_LOGS_DIR:/home/forest/logs":z \
+  -v "${DB_VOLUME}:/home/forest/forest_db" \
+  -v "${LOG_VOLUME}:/home/forest/logs" \
   --entrypoint /bin/bash \
   ghcr.io/chainsafe/forest:"${FOREST_TAG}" \
   -c "$COMMANDS" || exit 1
 
+# Dummy container to copy the snapshot files from the volume to the "host".
+COPIER=$(docker container create -v "${CHAIN_NAME}_db:/opt" busybox)
+docker run --rm -v "${DB_VOLUME}:/opt" busybox /bin/sh -c 'ls /opt/forest_snapshot_*.forest.car.zst' | while read -r file; do
+  docker cp "$COPIER":"$file" "$CHAIN_DB_DIR"
+done
+docker rm "$COPIER"
+
 aws --endpoint "$R2_ENDPOINT" s3 cp --no-progress "$CHAIN_DB_DIR/forest_snapshot_$CHAIN_NAME"*.forest.car.zst s3://"$SNAPSHOT_BUCKET"/"$CHAIN_NAME"/latest/ || exit 1
 
 # Delete snapshot files

tf-managed/modules/daily-snapshot/main.tf

@@ -7,7 +7,7 @@
 
 // Ugly hack because 'archive_file' cannot mix files and folders.
 data "external" "sources_tar" {
-  program = ["bash", "${path.module}/prep_sources.sh", path.module, var.common_resources_dir]
+  program = ["bash", "${path.module}/prep_sources.sh", path.module]
 }
 
 
@@ -32,30 +32,28 @@ data "digitalocean_ssh_keys" "keys" {
 
 # Set required environment variables
 locals {
-  env_content = templatefile("${path.module}/service/forest-env.tpl", {
-    R2_ACCESS_KEY        = var.R2_ACCESS_KEY,
-    R2_SECRET_KEY        = var.R2_SECRET_KEY,
-    r2_endpoint          = var.r2_endpoint,
-    slack_token          = var.slack_token,
-    slack_channel        = var.slack_channel,
-    snapshot_bucket      = var.snapshot_bucket,
-    snapshot_endpoint    = var.snapshot_endpoint,
-    NEW_RELIC_API_KEY    = var.new_relic_api_key,
-    NEW_RELIC_ACCOUNT_ID = var.new_relic_account_id,
-    NEW_RELIC_REGION     = var.new_relic_region,
-    BASE_FOLDER          = "/root",
-    forest_tag           = var.forest_tag
-  })
+  env_content = <<-EOT
+  R2_ACCESS_KEY=${var.R2_ACCESS_KEY}
+  R2_SECRET_KEY=${var.R2_SECRET_KEY}
+  R2_ENDPOINT=${var.r2_endpoint}
+  SNAPSHOT_BUCKET=${var.snapshot_bucket}
+  SLACK_API_TOKEN=${var.slack_token}
+  SLACK_NOTIFICATION_CHANNEL=${var.slack_channel}
+  FOREST_TAG=${var.forest_tag}
+  EOT
 }
 
 locals {
   init_commands = ["cd /root/",
     "tar xf sources.tar",
-    # Set required environment variables
     "echo '${local.env_content}' >> /root/.forest_env",
-    "echo '. ~/.forest_env' >> .bashrc",
-    ". ~/.forest_env",
-    "nohup sh ./init.sh > init_log.txt &",
+    <<-EOT
+    export NEW_RELIC_API_KEY=${var.new_relic_api_key}
+    export NEW_RELIC_ACCOUNT_ID=${var.new_relic_account_id}
+    export NEW_RELIC_REGION=${var.new_relic_region}
+    nohup sh ./init.sh > init_log.txt &
+    EOT
+    ,
     # Exiting without a sleep sometimes kills the script :-/
     "sleep 60s"
   ]

tf-managed/modules/daily-snapshot/prep_sources.sh

@@ -3,12 +3,8 @@
 # Enable strict error handling and command tracing
 set -euxo pipefail
 
-# Copy local source files in a folder together with ruby_common and create a zip archive.
-
+# Copy local source files in a folder, and create a zip archive.
 cd "$1"
-cp --archive "$2"/ruby_common service/
-
 rm -f sources.tar
 (cd service && tar cf ../sources.tar --sort=name --mtime='UTC 2019-01-01' ./* > /dev/null 2>&1)
-rm -fr service/ruby_common
 echo "{ \"path\": \"$1/sources.tar\" }"

tf-managed/modules/daily-snapshot/service/calibnet_cron_job

@@ -1,6 +1,5 @@
 #!/bin/bash
 
 # shellcheck source=/dev/null
-source ~/.forest_env
-cd "$BASE_FOLDER" || exit
-flock -n /tmp/calibnet.lock -c "ruby daily_snapshot.rb calibnet >> logs/calibnet_log.txt 2>&1"
+cd $HOME
+flock -n /tmp/calibnet.lock -c "docker run --privileged -v /var/run/docker.sock:/var/run/docker.sock --rm --env-file .forest_env -e NETWORK_CHAIN=calibnet ghcr.io/chainsafe/forest-snapshot-service:latest >> calibnet_log.txt 2>&1"

tf-managed/modules/daily-snapshot/service/init.sh

@@ -10,26 +10,9 @@ export DEBIAN_FRONTEND=noninteractive
 
 # Using timeout to ensure the script retries if the APT servers are temporarily unavailable.
 timeout 10m bash -c 'until apt-get -qqq --yes update && \
- apt-get -qqq --yes install ruby ruby-dev anacron awscli; do sleep 10; \
+ apt-get -qqq --yes install anacron ; do sleep 10; \
 done'
 
-# Install the gems
-gem install docker-api slack-ruby-client
-gem install activesupport -v 7.0.8
-
-# 1. Configure aws
-# 2. Create forest_db directory
-# 3. Copy scripts to /etc/cron.hourly
-
-## Configure aws
-aws configure set default.s3.multipart_chunksize 4GB
-aws configure set aws_access_key_id "$R2_ACCESS_KEY"
-aws configure set aws_secret_access_key "$R2_SECRET_KEY"
-
-## Create forest data directory
-mkdir forest_db logs
-chmod 777 forest_db logs
-
 # Run new_relic and fail2ban scripts
 bash newrelic_fail2ban.sh
 

tf-managed/modules/daily-snapshot/service/mainnet_cron_job

@@ -1,6 +1,5 @@
 #!/bin/bash
 
 # shellcheck source=/dev/null
-source ~/.forest_env
-cd "$BASE_FOLDER" || exit
-flock -n /tmp/mainnet.lock -c "ruby daily_snapshot.rb mainnet > mainnet_log.txt 2>&1" || exit
+cd $HOME
+flock -n /tmp/mainnet.lock -c "docker run --privileged -v /var/run/docker.sock:/var/run/docker.sock --rm --env-file .forest_env -e NETWORK_CHAIN=mainnet ghcr.io/chainsafe/forest-snapshot-service:latest >> mainnet_log.txt 2>&1"

tf-managed/modules/daily-snapshot/variable.tf

@@ -48,12 +48,6 @@ variable "r2_endpoint" {
   type        = string
 }
 
-variable "snapshot_endpoint" {
-  description = "S3 endpoint for the snapshots"
-  type        = string
-  default     = "https://fra1.digitaloceanspaces.com/"
-}
-
 variable "forest_tag" {
   description = "Image tag for the Forest container"
   type        = string
@@ -110,10 +104,6 @@ variable "new_relic_account_id" {
   sensitive   = true
 }
 
-variable "common_resources_dir" {
-  type = string
-}
-
 variable "environment" {
   description = "The environment name"
   type        = string