odysseus-core 0.9.0 → 0.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5fad8198958c2e3991a174b1d7821199e30c4807a48fe7bd9011f689ca430ae6
-  data.tar.gz: 9241a316d823c6397e3d293208bc186dcd0b6db88a3b1410e05599a43b99d686
+  metadata.gz: f6d1b931d263332fd0a431b4a0842949f52d78f2f366f8dbf72cb98f0ea95282
+  data.tar.gz: f3ea3a6477a88cbd38b1386f04e983bad5d7d782475f46322989568389e4fa5b
 SHA512:
-  metadata.gz: a323a1f3f08eded7052792ac13b83514a6fa045a7bcfb8bf928d28eb99ef4321ca8ec9abb2f9cd08e752467147e9cca8465bb60a3cc83d036c312cf3749c6a91
-  data.tar.gz: d800125c7690876145e8c7297d986d0a3ce1aba2856454df17567a209a7ae958eb79414afc0f4094390f82ad2205a61544cbc7ba31b583fd704040b15a0abcc4
+  metadata.gz: b89f611e20800ecfbdbd30fe571b6a9979877a59f52655a0c05fb3aeadc7567f6af4325711c92eb02698bbdd23c1746f6c1d150436e20dc45318a4f0fae0f1b4
+  data.tar.gz: 6db503709fb510a38980f0357a69f82f07b7b04e77b8b63c7286035f88425134039c06618164648876324a4a22f902341d4ca41fa4bb5a027a5addc07a4a7309

data/CHANGELOG.md

@@ -7,8 +7,31 @@ gem artifacts, so they are summaries rather than contemporaneous notes.
 
 ## [Unreleased]
 
+## [0.10.0] - 2026-08-28
+
 ### Fixed
 
+- Caddy no longer loses every route when its container is replaced or the host
+  reboots. Routes are added through the admin API and live in Caddy's memory;
+  Caddy autosaves them to `$XDG_CONFIG_HOME/caddy/autosave.json`, which the
+  official image sets to `/config` -- and odysseus mounted only `/data`
+  (`XDG_DATA_HOME`), so certificates persisted and routes were discarded with
+  the container. On a host with nine services, all nine stayed unreachable until
+  nine separate deploys ran. `HostPaths#caddy_config_dir` is now mounted at
+  `/config` and Caddy starts with `--resume`.
+
+  If `--resume` fails to start the container, odysseus renames the autosave to
+  `autosave.json.rejected` and starts once without it. `CADDY_IMAGE` is the
+  moving `caddy:2-alpine` tag, so a newer Caddy can meet an autosave it will not
+  parse, and `WebDeploy` aborts when Caddy does not come up -- unhandled, one
+  bad autosave would have failed every deploy on the host. The rename runs
+  inside a container: Caddy creates `/config/caddy` as root with mode 0700, so a
+  non-root deploy user cannot move the file from the host.
+
+  A host that has never run Caddy still has no routes until each service
+  deploys; `--resume` falls back to the image's Caddyfile there, which is what
+  creates `srv0`.
+
 - Commands echoed under `--debug`/`-v` no longer print secrets. `SSH#execute`
   echoed the command verbatim, and registry login is built as
   `echo '<password>' | docker login <server> -u <user> --password-stdin` --

data/lib/odysseus/caddy/client.rb

@@ -10,6 +10,11 @@ module Odysseus
       CONTAINER_NAME = 'odysseus-caddy'.freeze
       CADDY_IMAGE = 'caddy:2-alpine'.freeze
 
+      # The image's own CMD. Repeated here because --resume has to be appended
+      # to it: passing only --resume would drop the Caddyfile fallback that
+      # creates srv0 on a host with no autosave yet.
+      CADDY_RUN_CMD = 'caddy run --config /etc/caddy/Caddyfile --adapter caddyfile'.freeze
+
       # @param ssh [Odysseus::Deployer::SSH] SSH connection to server
       # @param docker [Odysseus::Docker::Client] Docker client
       def initialize(ssh:, docker:)
@@ -34,6 +39,17 @@ module Odysseus
       # Nothing is lost: certificates live in the mounted volume, and routes
       # are re-added by the deploy that follows.
       #
+      # Started with --resume, so the routes added through the admin API by
+      # previous deploys come back with the container rather than having to be
+      # re-added by redeploying every service on the host.
+      #
+      # If that start fails the autosave is the only new thing that could have
+      # caused it -- CADDY_IMAGE is a moving tag, so a newer Caddy can meet an
+      # autosave it will not parse. Since WebDeploy aborts when Caddy does not
+      # come up, leaving that unhandled would fail every deploy on the host
+      # until someone deleted the file by hand. The retry gives up the routes,
+      # which is exactly the behaviour that existed before --resume.
+      #
       # @return [Boolean] true if caddy is running
       def ensure_running
         return true if running?
@@ -41,6 +57,9 @@ module Odysseus
         @docker.remove(CONTAINER_NAME) if @docker.container_exists?(CONTAINER_NAME)
 
         start_caddy
+        return true if running?
+
+        restart_without_autosave
         running?
       end
 
@@ -51,12 +70,18 @@ module Odysseus
       end
 
       # Start Caddy container
-      def start_caddy
+      # @param resume [Boolean] reload the autosaved config (see #ensure_running)
+      def start_caddy(resume: true)
         # Create network if not exists (with label to protect from prune)
         @ssh.execute('docker network create --label odysseus.managed=true odysseus 2>/dev/null || true')
 
-        # Create data directory for certificates
+        # Create data directory for certificates, and the config directory
+        # holding the autosave. Two directories because the image sets
+        # XDG_DATA_HOME=/data and XDG_CONFIG_HOME=/config, and Caddy writes the
+        # autosave under the latter -- mounting only /data persisted the
+        # certificates and threw the routes away with the container.
         @ssh.execute("mkdir -p #{Shellwords.escape(host_paths.caddy_dir)}")
+        @ssh.execute("mkdir -p #{Shellwords.escape(host_paths.caddy_config_dir)}")
 
         # Run Caddy with admin API enabled and persistent storage for certs
         #
@@ -81,13 +106,17 @@ module Odysseus
             ports: ['80:80', '443:443', "127.0.0.1:#{ADMIN_API_PORT}:#{ADMIN_API_PORT}"],
             network: 'odysseus',
             restart: 'unless-stopped',
-            volumes: ["#{Shellwords.escape(host_paths.caddy_dir)}:/data"],
+            volumes: [
+              "#{Shellwords.escape(host_paths.caddy_dir)}:/data",
+              "#{Shellwords.escape(host_paths.caddy_config_dir)}:/config"
+            ],
             env: {
               'CADDY_ADMIN' => "0.0.0.0:#{ADMIN_API_PORT}"
             },
             labels: {
               'odysseus.managed' => 'true'
-            }
+            },
+            cmd: resume ? "#{CADDY_RUN_CMD} --resume" : CADDY_RUN_CMD
           }
         )
 
@@ -316,6 +345,29 @@ module Odysseus
 
       private
 
+      def restart_without_autosave
+        @docker.remove(CONTAINER_NAME) if @docker.container_exists?(CONTAINER_NAME)
+        discard_autosave
+        start_caddy(resume: false)
+      end
+
+      # Caddy creates /config/caddy as root with mode 0700, so a non-root deploy
+      # user cannot move the file from the host -- it has to be moved from
+      # inside a container. CADDY_IMAGE is guaranteed to be on the host: it is
+      # the image that just failed to start.
+      #
+      # Renamed rather than deleted, because if --resume was not the reason
+      # Caddy failed then this file is the only copy of the host's routes.
+      # Tolerates its own failure: a Caddy that died for some other reason may
+      # have no autosave at all, and that must not abort the retry.
+      def discard_autosave
+        mount = "#{Shellwords.escape(host_paths.caddy_config_dir)}:/config"
+        @ssh.execute(
+          "docker run --rm -v #{mount} #{CADDY_IMAGE} " \
+          'mv /config/caddy/autosave.json /config/caddy/autosave.json.rejected 2>/dev/null || true'
+        )
+      end
+
       def host_paths
         @host_paths ||= Odysseus::HostPaths.new(@ssh)
       end

data/lib/odysseus/core/version.rb

@@ -2,6 +2,6 @@
 
 module Odysseus
   module Core
-    VERSION = '0.9.0'
+    VERSION = '0.10.0'
   end
 end

data/lib/odysseus/host_paths.rb

@@ -57,6 +57,20 @@ module Odysseus
       File.join(base, 'caddy')
     end
 
+    # Caddy's /config mount: the autosave of its running configuration.
+    #
+    # The official image sets XDG_CONFIG_HOME=/config and XDG_DATA_HOME=/data,
+    # and Caddy writes $XDG_CONFIG_HOME/caddy/autosave.json on every admin API
+    # change. Mounting only #caddy_dir therefore persisted certificates and
+    # discarded the routes, which is why a removed or restarted Caddy came back
+    # serving nothing until every service on the host redeployed. Kept separate
+    # from #caddy_dir rather than nested inside it because the container owns
+    # the layout of both, and /data already has a meaning to Caddy.
+    # @return [String]
+    def caddy_config_dir
+      File.join(base, 'caddy-config')
+    end
+
     # Where a root install wrote, whoever is connected now. Used to read the
     # deploy history of a host that has since moved to a deploy user.
     # @return [String]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: odysseus-core
 version: !ruby/object:Gem::Version
-  version: 0.9.0
+  version: 0.10.0
 platform: ruby
 authors:
 - Thomas
@@ -157,7 +157,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.16
+rubygems_version: 4.0.19
 specification_version: 4
 summary: Core library for Odysseus deployment tool
 test_files: []