staging: test de Deuxfleurs/tricot#16

(fonctionnalité de tricot permettant de blocker certains user-agent)

.gitignore

@@ -4,3 +4,4 @@ secrets/*
 cluster/*/secrets/*
 !cluster/*/secrets/*.sample
 
+adrn-notes/

cluster/prod/app/backup/README.md

@@ -0,0 +1,32 @@
+## Pour remonter locement un backup de PSQL fait par Nomad (backup-weekly.hcl)
+
+```bash
+export AWS_BUCKET=backups-pgbasebackup
+export AWS_ENDPOINT=s3.deuxfleurs.shirokumo.net
+export AWS_ACCESS_KEY_ID=$(consul kv get "secrets/postgres/backup/aws_access_key_id")
+export AWS_SECRET_ACCESS_KEY=$(consul kv get secrets/postgres/backup/aws_secret_access_key)
+export CRYPT_PUBLIC_KEY=$(consul kv get secrets/postgres/backup/crypt_public_key)
+```
+
+Et voilà le travail : 
+
+```bash
+$ aws s3 --endpoint https://$AWS_ENDPOINT ls
+2022-04-14 17:00:50 backups-pgbasebackup
+
+$ aws s3 --endpoint https://$AWS_ENDPOINT ls s3://backups-pgbasebackup
+                           PRE 2024-07-28 00:00:36.140539/
+                           PRE 2024-08-04 00:00:21.291551/
+                           PRE 2024-08-11 00:00:26.589762/
+                           PRE 2024-08-18 00:00:40.873939/
+                           PRE 2024-08-25 01:03:54.672763/
+                           PRE 2024-09-01 00:00:20.019605/
+                           PRE 2024-09-08 00:00:16.969740/
+                           PRE 2024-09-15 00:00:37.951459/
+                           PRE 2024-09-22 00:00:21.030452/
+
+$ aws s3 --endpoint https://$AWS_ENDPOINT ls "s3://backups-pgbasebackup/2024-09-22 00:00:21.030452/"
+2024-09-22 03:23:28     623490 backup_manifest
+2024-09-22 03:25:32 6037121487 base.tar.gz
+2024-09-22 03:25:33   19948939 pg_wal.tar.gz
+```

cluster/prod/app/backup/build/backup-psql/backup-psql.py

@@ -44,6 +44,8 @@ if not client.bucket_exists(bucket):
   abort(f"Bucket {bucket} does not exist or its access is forbidden, aborting")
 
 # Perform the backup locally
+# Via command-line:
+# pg_basebackup --host=localhost --username=$PSQL_USER --pgdata=. --format=tar  --wal-method=stream --gzip --compress=6 --progress --max-rate=5M
 try:
   ret = subprocess.run(["pg_basebackup",
     f"--host={psql_host}",

cluster/prod/app/core/deploy/d53.hcl

@@ -1,5 +1,5 @@
 job "core-d53" {
-  datacenters = ["neptune", "scorpio", "bespin"]
+  datacenters = ["neptune", "scorpio", "bespin", "corrin"]
   type = "service"
   priority = 90
 

cluster/prod/app/core/deploy/diplonat.hcl

@@ -1,5 +1,5 @@
 job "core-diplonat" {
-  datacenters = ["neptune", "scorpio", "bespin"]
+  datacenters = ["neptune", "scorpio", "bespin", "corrin"]
   type = "system"
   priority = 90
 

cluster/prod/app/core/deploy/tricot.hcl

@@ -3,7 +3,7 @@ job "core-tricot" {
   # on pourra mettre bespin quand on aura migré gitea de la vm vers le cluster
   # en attendant, les deux  ne sont pas capables de partager les certificats SSL
   # donc on laisse la VM gitea gérer les certifs et prendre tout le trafic http(s)
-  datacenters = ["neptune", "scorpio"]
+  datacenters = ["corrin", "neptune", "scorpio"]
   type = "system"
   priority = 90
 
@@ -28,7 +28,7 @@ job "core-tricot" {
       driver = "docker"
 
       config {
-        image = "superboum/amd64_tricot:54"
+        image = "armael/tricot:n6dk1b5xrdww12zf12jbcmihqs6g1brz"
         network_mode = "host"
         readonly_rootfs = true
         ports = [ "http_port", "https_port" ]

cluster/prod/app/cryptpad/build/default.nix

@@ -71,16 +71,16 @@
   });
 in buildNpmPackage rec {
   pname = "cryptpad";
-  version = "2024.6.1";
+  version = "2024.9.0";
 
   src = fetchFromGitHub {
     owner = "cryptpad";
     repo = "cryptpad";
     rev = version;
-    hash = "sha256-qwyXpTY8Ds7R5687PVGZa/rlEyrAZjNzJ4+VQZpF8v0=";
+    hash = "sha256-OUtWaDVLRUbKS0apwY0aNq4MalGFv+fH9VA7LvWWYRs=";
   };
 
-  npmDepsHash = "sha256-GSTPsXqe/rxiDh5OW2t+ZY1YRNgRSDxkJ0pvcLIFtFw=";
+  npmDepsHash = "sha256-pK0b7q1kJja9l8ANwudbfo3jpldwuO56kuulS8X9A5s=";
 
   inherit nodejs;
 

cluster/prod/app/cryptpad/build/npins/sources.json

@@ -3,8 +3,8 @@
     "nixpkgs": {
       "type": "Channel",
       "name": "nixos-24.05",
-      "url": "https://releases.nixos.org/nixos/24.05/nixos-24.05.3311.a1cc729dcbc3/nixexprs.tar.xz",
-      "hash": "13al93fac4xdxj0jllfby2v9klwqdhsf3yg10mnsm9ys84v7gsnn"
+      "url": "https://releases.nixos.org/nixos/24.05/nixos-24.05.5385.1719f27dd95f/nixexprs.tar.xz",
+      "hash": "0f7i315g1z8kjh10hvj2zv7y2vfqxmwvd96hwlcrr8aig6qq5gzm"
     }
   },
   "version": 3

cluster/prod/app/cryptpad/build_docker/Dockerfile

@@ -0,0 +1,59 @@
+# SPDX-FileCopyrightText: 2023 XWiki CryptPad Team <contact@cryptpad.org> and contributors
+#
+# SPDX-License-Identifier: AGPL-3.0-or-later
+#
+# Tweaks by Deuxfleurs
+
+# Multistage build to reduce image size and increase security
+FROM node:lts-slim AS build
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && apt-get install --no-install-recommends -y \
+    ca-certificates tar wget
+
+# Download the release tarball
+RUN wget https://github.com/cryptpad/cryptpad/archive/refs/tags/2024.9.0.tar.gz -O cryptpad.tar.gz
+
+# Create folder for CryptPad
+RUN mkdir /cryptpad
+
+# Extract the release into /cryptpad
+RUN tar xvzf cryptpad.tar.gz -C /cryptpad --strip-components 1
+
+# Go to /cryptpad
+WORKDIR /cryptpad
+
+# Install dependencies
+RUN npm install --production && npm run install:components
+
+# Create the actual CryptPad image
+FROM node:lts-slim
+ENV DEBIAN_FRONTEND=noninteractive
+
+# Install curl for healthcheck
+# Install git, rdfind and unzip for install-onlyoffice.sh
+RUN apt-get update && apt-get install --no-install-recommends -y \
+    curl ca-certificates git rdfind unzip && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/*
+
+# Copy cryptpad with installed modules
+COPY --from=build /cryptpad /cryptpad
+
+# Set workdir to cryptpad
+WORKDIR /cryptpad
+
+# Install onlyoffice
+RUN ./install-onlyoffice.sh --accept-license --trust-repository
+
+# Build static pages (?) unsure we need this
+RUN npm run build
+
+# Healthcheck
+HEALTHCHECK --interval=1m CMD curl -f http://localhost:3000/ || exit 1
+
+# Ports
+EXPOSE 3000 3003
+
+# Run cryptpad on startup
+CMD ["npm", "start"]

cluster/prod/app/cryptpad/config/config-debug.js

@@ -0,0 +1,296 @@
+/* globals module */
+
+/*  DISCLAIMER:
+
+    There are two recommended methods of running a CryptPad instance:
+
+    1. Using a standalone nodejs server without HTTPS (suitable for local development)
+    2. Using NGINX to serve static assets and to handle HTTPS for API server's websocket traffic
+
+    We do not officially recommend or support Apache, Docker, Kubernetes, Traefik, or any other configuration.
+    Support requests for such setups should be directed to their authors.
+
+    If you're having difficulty difficulty configuring your instance
+    we suggest that you join the project's IRC/Matrix channel.
+
+    If you don't have any difficulty configuring your instance and you'd like to
+    support us for the work that went into making it pain-free we are quite happy
+    to accept donations via our opencollective page: https://opencollective.com/cryptpad
+
+*/
+module.exports = {
+/*  CryptPad is designed to serve its content over two domains.
+ *  Account passwords and cryptographic content is handled on the 'main' domain,
+ *  while the user interface is loaded on a 'sandbox' domain
+ *  which can only access information which the main domain willingly shares.
+ *
+ *  In the event of an XSS vulnerability in the UI (that's bad)
+ *  this system prevents attackers from gaining access to your account (that's good).
+ *
+ *  Most problems with new instances are related to this system blocking access
+ *  because of incorrectly configured sandboxes. If you only see a white screen
+ *  when you try to load CryptPad, this is probably the cause.
+ *
+ *  PLEASE READ THE FOLLOWING COMMENTS CAREFULLY.
+ *
+ */
+
+/*  httpUnsafeOrigin is the URL that clients will enter to load your instance.
+ *  Any other URL that somehow points to your instance is supposed to be blocked.
+ *  The default provided below assumes you are loading CryptPad from a server
+ *  which is running on the same machine, using port 3000.
+ *
+ *  In a production instance this should be available ONLY over HTTPS
+ *  using the default port for HTTPS (443) ie. https://cryptpad.fr
+ *  In such a case this should be also handled by NGINX, as documented in
+ *  cryptpad/docs/example.nginx.conf (see the $main_domain variable)
+ *
+ */
+    httpUnsafeOrigin: 'https://pad-debug.deuxfleurs.fr',
+
+/*  httpSafeOrigin is the URL that is used for the 'sandbox' described above.
+ *  If you're testing or developing with CryptPad on your local machine then
+ *  it is appropriate to leave this blank. The default behaviour is to serve
+ *  the main domain over port 3000 and to serve the sandbox content over port 3001.
+ *
+ *  This is not appropriate in a production environment where invasive networks
+ *  may filter traffic going over abnormal ports.
+ *  To correctly configure your production instance you must provide a URL
+ *  with a different domain (a subdomain is sufficient).
+ *  It will be used to load the UI in our 'sandbox' system.
+ *
+ *  This value corresponds to the $sandbox_domain variable
+ *  in the example nginx file.
+ *
+ *  Note that in order for the sandboxing system to be effective
+ *  httpSafeOrigin must be different from httpUnsafeOrigin.
+ *
+ *  CUSTOMIZE AND UNCOMMENT THIS FOR PRODUCTION INSTALLATIONS.
+ */
+    httpSafeOrigin: "https://pad-sandbox-debug.deuxfleurs.fr",
+
+/*  httpAddress specifies the address on which the nodejs server
+ *  should be accessible. By default it will listen on 127.0.0.1
+ *  (IPv4 localhost on most systems). If you want it to listen on
+ *  all addresses, including IPv6, set this to '::'.
+ *
+ */
+    httpAddress: '::',
+
+/*  httpPort specifies on which port the nodejs server should listen.
+ *  By default it will serve content over port 3000, which is suitable
+ *  for both local development and for use with the provided nginx example,
+ *  which will proxy websocket traffic to your node server.
+ *
+ */
+    httpPort: 3000,
+
+/*  httpSafePort allows you to specify an alternative port from which
+ *  the node process should serve sandboxed assets. The default value is
+ *  that of your httpPort + 1. You probably don't need to change this.
+ *
+ */
+    // httpSafePort: 3001,
+
+/*  CryptPad will launch a child process for every core available
+ *  in order to perform CPU-intensive tasks in parallel.
+ *  Some host environments may have a very large number of cores available
+ *  or you may want to limit how much computing power CryptPad can take.
+ *  If so, set 'maxWorkers' to a positive integer.
+ */
+    // maxWorkers: 4,
+
+    /* =====================
+     *         Admin
+     * ===================== */
+
+    /*
+     *  CryptPad contains an administration panel. Its access is restricted to specific
+     *  users using the following list.
+     *  To give access to the admin panel to a user account, just add their public signing
+     *  key, which can be found on the settings page for registered users.
+     *  Entries should be strings separated by a comma.
+     */
+    adminKeys: [
+      "[quentin@pad.deuxfleurs.fr/EWtzm-CiqJnM9RZL9mj-YyTgAtX-Zh76sru1K5bFpN8=]",
+      "[adrn@pad.deuxfleurs.fr/PxDpkPwd-jDJWkfWdAzFX7wtnLpnPlBeYZ4MmoEYS6E=]",
+      "[lx@pad.deuxfleurs.fr/FwQzcXywx1FIb83z6COB7c3sHnz8rNSDX1xhjPuH3Fg=]",
+      "[trinity-1686a@pad-debug.deuxfleurs.fr/Pu6Ef03jEsAGBbZI6IOdKd6+5pORD5N51QIYt4-Ys1c=]",
+      "[Jill@pad.deuxfleurs.fr/tLW7W8EVNB2KYETXEaOYR+HmNiBQtZj7u+SOxS3hGmg=]",
+      "[vincent@pad.deuxfleurs.fr/07FQiE8w1iztRWwzbRJzEy3xIqnNR31mUFjLNiGXjwU=]",
+      "[boris@pad.deuxfleurs.fr/kHo5LIhSxDFk39GuhGRp+XKlMjNe+lWfFWM75cINoTQ=]",
+      "[maximilien@pad.deuxfleurs.fr/UoXHLejYRUjvX6t55hAQKpjMdU-3ecg4eDhAeckZmyE=]",
+      "[armael@pad-debug.deuxfleurs.fr/CIKMvNdFxGavwTmni0TnR3x9GM0ypgx3DMcFyzppplU=]",
+      "[bjonglez@pad-debug.deuxfleurs.fr/+RRzwcLPj5ZCWELUXMjmt3u+-lvYnyhpDt4cqAn9nh8=]"
+    ],
+
+    /* =====================
+     *        STORAGE
+     * ===================== */
+
+    /*  Pads that are not 'pinned' by any registered user can be set to expire
+     *  after a configurable number of days of inactivity (default 90 days).
+     *  The value can be changed or set to false to remove expiration.
+     *  Expired pads can then be removed using a cron job calling the
+     *  `evict-inactive.js` script with node
+     *
+     *  defaults to 90 days if nothing is provided
+     */
+    //inactiveTime: 90, // days
+
+    /*  CryptPad archives some data instead of deleting it outright.
+     *  This archived data still takes up space and so you'll probably still want to
+     *  remove these files after a brief period.
+     *
+     *  cryptpad/scripts/evict-inactive.js is intended to be run daily
+     *  from a crontab or similar scheduling service.
+     *
+     *  The intent with this feature is to provide a safety net in case of accidental
+     *  deletion. Set this value to the number of days you'd like to retain
+     *  archived data before it's removed permanently.
+     *
+     *  defaults to 15 days if nothing is provided
+     */
+    //archiveRetentionTime: 15,
+
+    /*  It's possible to configure your instance to remove data
+     *  stored on behalf of inactive accounts. Set 'accountRetentionTime'
+     *  to the number of days an account can remain idle before its
+     *  documents and other account data is removed.
+     *
+     *  Leave this value commented out to preserve all data stored
+     *  by user accounts regardless of inactivity.
+     */
+     //accountRetentionTime: 365,
+
+    /*  Starting with CryptPad 3.23.0, the server automatically runs
+     *  the script responsible for removing inactive data according to
+     *  your configured definition of inactivity. Set this value to `true`
+     *  if you prefer not to remove inactive data, or if you prefer to
+     *  do so manually using `scripts/evict-inactive.js`.
+     */
+    //disableIntegratedEviction: true,
+
+
+    /*  Max Upload Size (bytes)
+     *  this sets the maximum size of any one file uploaded to the server.
+     *  anything larger than this size will be rejected
+     *  defaults to 20MB if no value is provided
+     */
+    //maxUploadSize: 20 * 1024 * 1024,
+
+    /*  Users with premium accounts (those with a plan included in their customLimit)
+     *  can benefit from an increased upload size limit. By default they are restricted to the same
+     *  upload size as any other registered user.
+     *
+     */
+    //premiumUploadSize: 100 * 1024 * 1024,
+
+    /* =====================
+     *   DATABASE VOLUMES
+     * ===================== */
+
+    /*
+     * We need this config entry, else CryptPad will try to mkdir
+     * some stuff into Nix store apparently...
+     */
+    base: '/mnt/data',
+
+    /*
+     *  CryptPad stores each document in an individual file on your hard drive.
+     *  Specify a directory where files should be stored.
+     *  It will be created automatically if it does not already exist.
+     */
+    filePath: '/mnt/datastore/',
+
+    /*  CryptPad offers the ability to archive data for a configurable period
+     *  before deleting it, allowing a means of recovering data in the event
+     *  that it was deleted accidentally.
+     *
+     *  To set the location of this archive directory to a custom value, change
+     *  the path below:
+     */
+    archivePath: '/mnt/data/archive',
+
+    /*  CryptPad allows logged in users to request that particular documents be
+     *  stored by the server indefinitely. This is called 'pinning'.
+     *  Pin requests are stored in a pin-store. The location of this store is
+     *  defined here.
+     */
+    pinPath: '/mnt/data/pins',
+
+    /*  if you would like the list of scheduled tasks to be stored in
+        a custom location, change the path below:
+    */
+    taskPath: '/mnt/data/tasks',
+
+    /*  if you would like users' authenticated blocks to be stored in
+        a custom location, change the path below:
+    */
+    blockPath: '/mnt/block',
+
+    /*  CryptPad allows logged in users to upload encrypted files. Files/blobs
+     *  are stored in a 'blob-store'. Set its location here.
+     */
+    blobPath: '/mnt/blob',
+
+    /*  CryptPad stores incomplete blobs in a 'staging' area until they are
+     *  fully uploaded. Set its location here.
+     */
+    blobStagingPath: '/mnt/data/blobstage',
+
+    decreePath: '/mnt/data/decrees',
+
+    /* CryptPad supports logging events directly to the disk in a 'logs' directory
+     * Set its location here, or set it to false (or nothing) if you'd rather not log
+     */
+    logPath: false,
+
+    /* =====================
+     *       Debugging
+     * ===================== */
+
+    /*  CryptPad can log activity to stdout
+     *  This may be useful for debugging
+     */
+    logToStdout: true,
+
+    /* CryptPad can be configured to log more or less
+     * the various settings are listed below by order of importance
+     *
+     * silly, verbose, debug, feedback, info, warn, error
+     *
+     * Choose the least important level of logging you wish to see.
+     * For example, a 'silly' logLevel will display everything,
+     * while 'info' will display 'info', 'warn', and 'error' logs
+     *
+     * This will affect both logging to the console and the disk.
+     */
+    logLevel: 'silly',
+
+    /*  clients can use the /settings/ app to opt out of usage feedback
+     *  which informs the server of things like how much each app is being
+     *  used, and whether certain clientside features are supported by
+     *  the client's browser. The intent is to provide feedback to the admin
+     *  such that the service can be improved. Enable this with `true`
+     *  and ignore feedback with `false` or by commenting the attribute
+     *
+     *  You will need to set your logLevel to include 'feedback'. Set this
+     *  to false if you'd like to exclude feedback from your logs.
+     */
+    logFeedback: false,
+
+    /*  CryptPad supports verbose logging
+     *  (false by default)
+     */
+    verbose: true,
+
+    /*  Surplus information:
+     *
+     *  'installMethod' is included in server telemetry to voluntarily
+     *  indicate how many instances are using unofficial installation methods
+     *  such as Docker.
+     *
+     */
+    installMethod: 'deuxfleurs.fr',
+};

cluster/prod/app/cryptpad/config/config.js

@@ -119,7 +119,9 @@ module.exports = {
       "[Jill@pad.deuxfleurs.fr/tLW7W8EVNB2KYETXEaOYR+HmNiBQtZj7u+SOxS3hGmg=]",
       "[vincent@pad.deuxfleurs.fr/07FQiE8w1iztRWwzbRJzEy3xIqnNR31mUFjLNiGXjwU=]",
       "[boris@pad.deuxfleurs.fr/kHo5LIhSxDFk39GuhGRp+XKlMjNe+lWfFWM75cINoTQ=]",
-      "[maximilien@pad.deuxfleurs.fr/UoXHLejYRUjvX6t55hAQKpjMdU-3ecg4eDhAeckZmyE=]"
+      "[maximilien@pad.deuxfleurs.fr/UoXHLejYRUjvX6t55hAQKpjMdU-3ecg4eDhAeckZmyE=]",
+      "[armael@pad.deuxfleurs.fr/CIKMvNdFxGavwTmni0TnR3x9GM0ypgx3DMcFyzppplU=]",
+      "[bjonglez@pad.deuxfleurs.fr/+RRzwcLPj5ZCWELUXMjmt3u+-lvYnyhpDt4cqAn9nh8=]"
     ],
 
     /* =====================

cluster/prod/app/cryptpad/deploy/cryptpad-debug.hcl

@@ -0,0 +1,80 @@
+job "cryptpad-debug" {
+  datacenters = ["neptune"]
+  type = "service"
+
+  group "cryptpad" {
+    count = 1
+
+    network {
+      port "http" {
+        to = 3000
+      }
+    }
+
+    restart {
+      attempts = 10
+      delay    = "30s"
+    }
+
+    task "main" {
+      driver = "docker"
+
+      constraint {
+        attribute = "${attr.unique.hostname}"
+        operator = "="
+        value = "courgette"
+      }
+
+      config {
+        image = "armael/cryptpad:2024.9.0"
+        ports = [ "http" ]
+
+        volumes = [
+          "/mnt/ssd/cryptpad-debug:/mnt",
+          "secrets/config-debug.js:/cryptpad/config.js",
+        ]
+      }
+      env {
+        CRYPTPAD_CONFIG = "/cryptpad/config.js"
+      }
+
+      template {
+        data = file("../config/config-debug.js")
+        destination = "secrets/config-debug.js"
+      }
+
+      /* Disabled because it requires modifications to the docker image and I do not want to invest the time yet
+      template {
+        data = file("../config/application_config-debug.js")
+        destination = "secrets/config-debug.js"
+      }
+      */
+
+      resources {
+        memory = 1000
+        cpu = 500
+      }
+
+      service {
+        name = "cryptpad-debug"
+        port = "http"
+        tags = [
+          "tricot pad-debug.deuxfleurs.fr",
+          "tricot pad-sandbox-debug.deuxfleurs.fr",
+          "tricot-add-header Cross-Origin-Resource-Policy cross-origin",
+          "tricot-add-header Cross-Origin-Embedder-Policy require-corp",
+          "tricot-add-header Access-Control-Allow-Origin *",
+          "tricot-add-header Access-Control-Allow-Credentials true",
+          "d53-cname pad-debug.deuxfleurs.fr",
+          "d53-cname pad-sandbox-debug.deuxfleurs.fr",
+        ]
+        check {
+          type = "http"
+          path = "/"
+          interval = "10s"
+          timeout = "2s"
+        }
+      }
+    }
+  }
+}

cluster/prod/app/cryptpad/deploy/cryptpad.hcl

@@ -26,7 +26,7 @@ job "cryptpad" {
       }
 
       config {
-        image = "kokakiwi/cryptpad:2024.6.1"
+        image = "kokakiwi/cryptpad:2024.9.0"
         ports = [ "http" ]
 
         volumes = [
@@ -63,6 +63,8 @@ job "cryptpad" {
           "tricot pad-sandbox.deuxfleurs.fr",
           "tricot-add-header Cross-Origin-Resource-Policy cross-origin",
           "tricot-add-header Cross-Origin-Embedder-Policy require-corp",
+          "tricot-add-header Access-Control-Allow-Origin *",
+          "tricot-add-header Access-Control-Allow-Credentials true",
           "d53-cname pad.deuxfleurs.fr",
           "d53-cname pad-sandbox.deuxfleurs.fr",
         ]

cluster/prod/app/email/config/postfix/main.cf

@@ -83,11 +83,14 @@ smtpd_forbid_unauth_pipelining = yes
 smtpd_discard_ehlo_keywords = chunking
 smtpd_forbid_bare_newline = yes
 
-smtpd_client_connection_rate_limit = 2
-
 #===
 # Rate limiting
 #===
+smtpd_client_connection_rate_limit = 2
+# do not rate-limit ourselves
+# in particular, useful for forgejo who opens a lot of SMTP connections
+smtpd_client_event_limit_exceptions = $mynetworks /etc/postfix/rate-limit-exceptions
+
 slow_destination_recipient_limit = 20
 slow_destination_concurrency_limit = 2
 

cluster/prod/app/email/deploy/email-android7.hcl

@@ -1,5 +1,6 @@
 job "email-android7" {
-  datacenters = ["neptune", "bespin"]
+  # Should not run on the same site as email.hcl (port conflict in diplonat)
+  datacenters = ["scorpio", "bespin"]
   type = "service"
   priority = 100
 
@@ -41,7 +42,7 @@ job "email-android7" {
 
       resources {
         cpu = 50
-        memory = 50
+        memory = 200
       }
 
       service {
@@ -95,7 +96,7 @@ job "email-android7" {
 
       resources {
         cpu = 50
-        memory = 50
+        memory = 200
       }
 
       service {

cluster/prod/app/email/deploy/email.hcl

@@ -1,4 +1,5 @@
 job "email" {
+  # Should not run on the same site as email-android7.hcl (port conflict in diplonat)
   datacenters = ["neptune"]
   type = "service"
   priority = 65
@@ -381,6 +382,29 @@ job "email" {
         destination = "secrets/postfix/transport"
       }
 
+      template {
+        # Collect machine IPs from the cluster.
+        # We use intermediate maps to ensure we get a sorted list with no duplicates,
+        # so that it is robust wrt. changes in the order of the output of ls or
+        # addition of new machines in an existing site.
+        # (scratch.MapValues returns the list of *values* in the map, sorted by *key*)
+        data = <<EOH
+        {{- range ls "diplonat/autodiscovery/ipv4" }}
+          {{- with $a := .Value | parseJSON }}
+            {{- scratch.MapSet "ipv4" $a.address $a.address }}
+          {{- end }}
+        {{- end -}}
+        {{- range ls "diplonat/autodiscovery/ipv6" }}
+          {{- with $a := .Value | parseJSON }}
+            {{- scratch.MapSet "ipv6" $a.address $a.address }}
+          {{- end }}
+        {{- end -}}
+        {{- range scratch.MapValues "ipv4" }}{{ . }} {{ end }}
+        {{- range scratch.MapValues "ipv6" }}[{{ . }}] {{ end }}
+        EOH
+        destination = "secrets/postfix/rate-limit-exceptions"
+      }
+
       # --- secrets ---
       template {
         data = "{{ with $d := key \"tricot/certs/smtp.deuxfleurs.fr\" | parseJSON }}{{ $d.cert_pem }}{{ end }}"

cluster/prod/app/guichet/deploy/guichet.hcl

@@ -28,7 +28,11 @@ job "guichet" {
       }
 
       resources {
-        memory = 200
+        # limite de mémoire un peu élevée par précaution.
+        # avec 200M, j'ai observé guichet se faire OOM-killed au moment
+        # où un nouvel utilisateur clique sur un lien d'invitation
+        # fraichement généré.
+        memory = 300
       }
 
       service {

cluster/prod/app/jitsi/build/docker-compose.yml

@@ -6,16 +6,17 @@ services:
       context: ./jitsi-meet
       args:
         # https://github.com/jitsi/jitsi-meet
-        MEET_TAG: stable/jitsi-meet_8252
-    image: superboum/amd64_jitsi_meet:v6
+        MEET_TAG: stable/jitsi-meet_9646
+        NODE_MAJOR_VERSION: 22
+    image: superboum/amd64_jitsi_meet:v7
 
   jitsi-conference-focus:
     build:
       context: ./jitsi-conference-focus
       args:
         # https://github.com/jitsi/jicofo
-        JICOFO_TAG: stable/jitsi-meet_8252
-    image: superboum/amd64_jitsi_conference_focus:v10
+        JICOFO_TAG: stable/jitsi-meet_9646
+    image: superboum/amd64_jitsi_conference_focus:v11
 
   jitsi-videobridge:
     build:
@@ -23,13 +24,13 @@ services:
       args:
         # https://github.com/jitsi/jitsi-videobridge
         # note: JVB is not tagged with non-stable tags
-        JVB_TAG: stable/jitsi-meet_8252
-    image: superboum/amd64_jitsi_videobridge:v21
+        JVB_TAG: stable/jitsi-meet_9646
+    image: superboum/amd64_jitsi_videobridge:v22
 
   jitsi-xmpp:
     build:
       context: ./jitsi-xmpp
       args:
-        MEET_TAG: stable/jitsi-meet_8252
-        PROSODY_VERSION: 1nightly191-1~bookworm
-    image: superboum/amd64_jitsi_xmpp:v11
+        MEET_TAG: stable/jitsi-meet_9646
+        PROSODY_VERSION: 0.12.3-1
+    image: superboum/amd64_jitsi_xmpp:v12

cluster/prod/app/jitsi/build/jitsi-meet/Dockerfile

@@ -1,8 +1,9 @@
 FROM debian:bookworm AS builder
 
+ARG NODE_MAJOR_VERSION
 RUN apt-get update && \
     apt-get install -y curl && \
-    curl -sL https://deb.nodesource.com/setup_19.x | bash - && \
+    curl -sL https://deb.nodesource.com/setup_${NODE_MAJOR_VERSION}.x | bash - && \
     apt-get install -y git nodejs make git unzip
 
 ARG MEET_TAG

cluster/prod/app/jitsi/build/jitsi-videobridge/jvb_run

@@ -6,7 +6,7 @@ if [ -z "${JITSI_NAT_LOCAL_IP}" ]; then
 fi
 
 if [ -z "${JITSI_NAT_PUBLIC_IP}" ]; then
-  JITSI_NAT_PUBLIC_IP=$(curl https://ifconfig.me)
+  JITSI_NAT_PUBLIC_IP=$(curl -4 https://ifconfig.me)
 fi
 
 echo "NAT config: ${JITSI_NAT_LOCAL_IP} -> ${JITSI_NAT_PUBLIC_IP}"

cluster/prod/app/jitsi/build/jitsi-xmpp/Dockerfile

@@ -13,8 +13,8 @@ RUN apt-get update && \
     apt-get install -y wget gnupg2 extrepo && \
     extrepo enable prosody && \
     apt-get update && \
-    apt-cache show prosody-0.12 && \
-    apt-get install -y prosody-0.12=${PROSODY_VERSION} lua-event
+    apt-cache show prosody && \
+    apt-get install -y prosody=${PROSODY_VERSION} lua-event
 
 RUN mkdir -p /usr/local/share/ca-certificates/ && \
     ln -sf \

cluster/prod/app/jitsi/config/config.js

@@ -369,7 +369,7 @@ var config = {
     // Message to show the users. Example: 'The service will be down for
     // maintenance at 01:00 AM GMT,
     // Does only support plaintext. No line skip.
-    noticeMessage: "Suite à une utilisation contraire à nos CGU, Deuxfleurs surveille activement cette instance Jitsi et enverra tout contenu illégal à la police. Pour toute question, commentaire ou suggestion, contactez moderation@deuxfleurs.fr . Following usage breaching our TOS, Deuxfleurs actively monitors this Jitsi instance and will send any illegal behavior to the Police. For any question, remark or suggestion, reach moderation@deuxfleurs.fr",
+    // noticeMessage: "Suite à une utilisation contraire à nos CGU, Deuxfleurs surveille activement cette instance Jitsi et enverra tout contenu illégal à la police. Pour toute question, commentaire ou suggestion, contactez moderation@deuxfleurs.fr . Following usage breaching our TOS, Deuxfleurs actively monitors this Jitsi instance and will send any illegal behavior to the Police. For any question, remark or suggestion, reach moderation@deuxfleurs.fr",
 
     // Enables calendar integration, depends on googleApiApplicationClientID
     // and microsoftApiApplicationClientID

cluster/prod/app/jitsi/config/videobridge.conf

@@ -115,7 +115,8 @@ videobridge {
     # (e.g. health or debug stats)
     private {
       # See JettyBundleActivatorConfig in Jicoco for values
-      host = 127.0.0.1
+      host = 0.0.0.0
+      port = {{ env "NOMAD_PORT_management_port" }}
     }
   }
   octo {

cluster/prod/app/jitsi/deploy/jitsi.hcl

@@ -1,5 +1,5 @@
 job "jitsi" {
-  datacenters = ["neptune", "scorpio"]
+  datacenters = ["neptune", "scorpio", "corrin"]
   type = "service"
 
   priority = 50
@@ -20,7 +20,7 @@ job "jitsi" {
     task "xmpp" {
       driver = "docker"
       config {
-        image = "superboum/amd64_jitsi_xmpp:v11"
+        image = "superboum/amd64_jitsi_xmpp:v12"
         ports = [ "bosh_port", "xmpp_port" ]
         network_mode = "host"
         volumes = [
@@ -101,7 +101,7 @@ EOF
     task "front" {
       driver = "docker"
       config {
-        image = "superboum/amd64_jitsi_meet:v6"
+        image = "superboum/amd64_jitsi_meet:v7"
         network_mode = "host"
         ports = [ "https_port" ]
         volumes = [
@@ -168,7 +168,7 @@ EOF
     task "jicofo" {
       driver = "docker"
       config {
-        image = "superboum/amd64_jitsi_conference_focus:v10"
+        image = "superboum/amd64_jitsi_conference_focus:v11"
         network_mode = "host"
         volumes = [
           "secrets/certs/jitsi.crt:/usr/local/share/ca-certificates/jitsi.crt",
@@ -203,14 +203,15 @@ EOF
   group "data_plane" {
     network {
       port "video_port" { static = 8080 }
+      port "management_port" { static = 8000 }
     }
 
     task "videobridge" {
       driver = "docker"
       config {
-        image = "superboum/amd64_jitsi_videobridge:v21"
+        image = "superboum/amd64_jitsi_videobridge:v22"
         network_mode = "host"
-        ports = [ "video_port" ] 
+        ports = [ "video_port", "management_port" ] 
         ulimit {
           nofile = "1048576:1048576"
           nproc = "65536:65536"
@@ -259,9 +260,16 @@ EOF
         port = "video_port"
         address_mode = "host"
         name = "video-jitsi"
+      }
+
+      service {
+        tags = [ "jitsi" ]
+        port = "management_port"
+        address_mode = "host"
+        name = "management-video-jitsi"
         check {
           type = "tcp"
-          port = "video_port"
+          port = "management_port"
           interval = "60s"
           timeout = "5s"
         }

cluster/prod/app/jitsi/integration/README.md

@@ -52,7 +52,7 @@ But maybe this value is deprecated: the check is still here but it is not used a
 start a maintainance container
 
 ```
-docker run --rm -it -v `pwd`/prosody/certs/:/var/lib/prosody/ -v `pwd`/prosody/prosody.cfg.lua:/etc/prosody/prosody.cfg.lua:ro --user root superboum/amd64_jitsi_xmpp:v11 bash
+docker run --rm -it -v `pwd`/prosody/certs/:/var/lib/prosody/ -v `pwd`/prosody/prosody.cfg.lua:/etc/prosody/prosody.cfg.lua:ro --user root superboum/amd64_jitsi_xmpp:v12 bash
 ```
 
 then generate certificates from inside this container

cluster/prod/app/jitsi/integration/docker-compose.yml

@@ -1,7 +1,7 @@
 version: '3.4'
 services:
   jitsi-xmpp:
-    image: superboum/amd64_jitsi_xmpp:v11
+    image: superboum/amd64_jitsi_xmpp:v12
     volumes:
       - "./prosody/prosody.cfg.lua:/etc/prosody/prosody.cfg.lua:ro"
       - "./prosody/certs/jitsi.crt:/var/lib/prosody/jitsi.crt:ro"
@@ -11,16 +11,19 @@ services:
     environment:
       - JICOFO_AUTH_PASSWORD=jicofopass
       - JVB_AUTH_PASSWORD=jvbpass
+    ports:
+      - "5222:5222/tcp"
  
   jitsi-conference-focus:
-    image: superboum/amd64_jitsi_conference_focus:v10
+    image: superboum/amd64_jitsi_conference_focus:v11
     volumes:
       - "./prosody/certs/jitsi.crt:/usr/local/share/ca-certificates/jitsi.crt:ro"
       - "./prosody/certs/auth.jitsi.crt:/usr/local/share/ca-certificates/auth.jitsi.crt:ro"
       - "./jicofo/jicofo.conf:/etc/jitsi/jicofo.conf:ro"
 
   jitsi-videobridge:
-    image: superboum/amd64_jitsi_videobridge:v21
+    image: superboum/amd64_jitsi_videobridge:v22
+    network_mode: "host"
     volumes:
       - "./prosody/certs/jitsi.crt:/usr/local/share/ca-certificates/jitsi.crt:ro"
       - "./prosody/certs/auth.jitsi.crt:/usr/local/share/ca-certificates/auth.jitsi.crt:ro"
@@ -31,7 +34,7 @@ services:
       - "10000:10000/udp"
 
   jitsi-meet:
-    image: superboum/amd64_jitsi_meet:v6
+    image: superboum/amd64_jitsi_meet:v7
     volumes:
       - "./prosody/certs/jitsi.crt:/etc/nginx/jitsi.crt:ro"
       - "./prosody/certs/jitsi.key:/etc/nginx/jitsi.key:ro"

cluster/prod/app/jitsi/integration/jvb/videobridge.conf

@@ -62,7 +62,7 @@ videobridge {
 
       configs {
         unique-xmpp-server {
-          hostname="jitsi-xmpp"
+          hostname="172.17.0.1"
           domain = "auth.jitsi"
           username = "jvb"
           password = "jvbpass"

cluster/prod/app/jitsi/integration/meet/config.js

@@ -22,7 +22,7 @@ var config = {
     },
 
     // BOSH URL. FIXME: use XEP-0156 to discover it.
-    bosh: '//192.168.1.143/http-bind',
+    bosh: '//[2a0c:e303:0:2a00::de6]/http-bind',
 
     // Websocket URL
     // websocket: 'wss://jitsi-meet.example.com/xmpp-websocket',

cluster/prod/app/matrix/README.md

@@ -0,0 +1,14 @@
+# Informations relatives à la config Matrix
+
+## Ressources
+
+- La doc de Synapse est là : https://element-hq.github.io/synapse/latest/welcome_and_overview.html
+
+### Métriques 
+
+- La page pour configurer les metrics : https://element-hq.github.io/synapse/latest/usage/configuration/config_documentation.html?highlight=metrics#metrics
+- La page pour le tutoriel sur configurer les metrics avec Prometheus : https://element-hq.github.io/synapse/latest/metrics-howto.html?highlight=metrics#how-to-monitor-synapse-metrics-using-prometheus
+
+---
+
+> Avec Nix on n'aurait pas tous ces problèmes.

cluster/prod/app/matrix/build/docker-compose.yml

@@ -6,18 +6,18 @@ services:
       context: ./riotweb
       args:
         # https://github.com/vector-im/element-web/releases
-        VERSION: 1.11.49
-    image: lxpz/amd64_elementweb:v35
+        VERSION: v1.11.78
+    image: particallydone/amd64_elementweb:v36
 
   synapse:
     build:
       context: ./matrix-synapse
       args:
         # https://github.com/matrix-org/synapse/releases
-        VERSION: 1.95.1
+        VERSION: v1.104.0
         # https://github.com/matrix-org/synapse-s3-storage-provider/commits/main
         # Update with the latest commit on main each time you update the synapse version
         # otherwise synapse may fail to launch due to incompatibility issues
         # see this issue for an example: https://github.com/matrix-org/synapse-s3-storage-provider/issues/64
-        S3_VERSION: v1.2.1
-    image: lxpz/amd64_synapse:v58
+        S3_VERSION: 2c46a764f700e6439afa11c00db827ddf21a9e89
+    image: particallydone/amd64_synapse:v60

cluster/prod/app/matrix/build/matrix-synapse/Dockerfile

@@ -1,4 +1,4 @@
-FROM amd64/debian:bookworm as builder
+FROM amd64/debian:trixie AS builder
 
 ARG VERSION
 ARG S3_VERSION
@@ -22,21 +22,25 @@ RUN apt-get update && \
       libpq-dev \
       virtualenv \
       libxslt1-dev \
-      git && \
-    virtualenv /root/matrix-env -p /usr/bin/python3 && \
+      git
+
+RUN virtualenv /root/matrix-env -p /usr/bin/python3 && \
     . /root/matrix-env/bin/activate && \
     pip3 install \
-      https://github.com/matrix-org/synapse/archive/v${VERSION}.tar.gz#egg=matrix-synapse[matrix-synapse-ldap3,postgres,resources.consent,saml2,url_preview] && \
+      https://github.com/element-hq/synapse/archive/${VERSION}.tar.gz#egg=matrix-synapse[matrix-synapse-ldap3,postgres,resources.consent,saml2,url_preview] && \
     pip3 install \
       git+https://github.com/matrix-org/synapse-s3-storage-provider.git@${S3_VERSION}
 
-FROM amd64/debian:bookworm
+# WARNING: trixie n'est pas une LTS
+# mais on est obligé d'avoir la même version que le builder
+# et le builder veut une version de rustc qui n'est pas dans bookworm (dernière LTS at the time of writing)
+FROM amd64/debian:trixie
 
 RUN apt-get update && \
     apt-get -qq -y full-upgrade && \
     apt-get install -y \
       python3 \
-      python3-distutils \
+      python3-setuptools \
       libffi8 \
       libjpeg62-turbo \
       libssl3 \

cluster/prod/app/matrix/build/riotweb/Dockerfile

@@ -1,13 +1,16 @@
-FROM amd64/debian:buster as builder
+FROM amd64/debian:trixie as builder
 
 ARG VERSION
 WORKDIR /root
 
 RUN apt-get update && \
-    apt-get install -y wget && \
-    wget https://github.com/vector-im/element-web/releases/download/v${VERSION}/element-v${VERSION}.tar.gz && \
-    tar xf element-v${VERSION}.tar.gz && \
-    mv element-v${VERSION}/ riot/
+    apt-get install -y wget
 
+RUN wget https://github.com/element-hq/element-web/releases/download/${VERSION}/element-${VERSION}.tar.gz && \
+    tar xf element-${VERSION}.tar.gz && \
+    mv element-${VERSION}/ riot/
+
+# Le conteneur de superboum contient uniquement un serveur web de 5 lignes.
+# Ca vous ennuie ? On peut publier Riot dans un bucket web Garage, tkt, ça sera Tricot qui servira.
 FROM superboum/amd64_webserver:v3
 COPY --from=builder /root/riot /srv/http

cluster/prod/app/matrix/deploy/im.hcl

@@ -15,7 +15,7 @@ job "matrix" {
       driver = "docker"
 
       config {
-        image = "lxpz/amd64_synapse:v58"
+        image = "particallydone/amd64_synapse:v60"
         network_mode = "host"
         readonly_rootfs = true
         ports = [ "api_port" ]
@@ -101,7 +101,7 @@ job "matrix" {
       driver = "docker"
 
       config {
-        image = "lxpz/amd64_synapse:v58"
+        image = "particallydone/amd64_synapse:v60"
         readonly_rootfs = true
         command = "/usr/local/bin/matrix-s3-async"
         work_dir = "/tmp"
@@ -137,7 +137,7 @@ EOH
     task "riotweb" {
       driver = "docker"
       config {
-        image = "lxpz/amd64_elementweb:v35"
+        image = "particallydone/amd64_elementweb:v36"
         ports = [ "web_port" ]
         volumes = [
           "secrets/config.json:/srv/http/config.json"
@@ -190,7 +190,7 @@ EOH
       driver = "docker"
 
       config {
-        image = "ghcr.io/matrix-org/sliding-sync:v0.99.12"
+        image = "ghcr.io/matrix-org/sliding-sync:v0.99.16"
         ports = [ "syncv3_api", "syncv3_metrics" ]
       }
 

cluster/prod/app/telemetry/config/prometheus.yml

@@ -26,6 +26,16 @@ scrape_configs:
         cert_file: /etc/prometheus/consul-client.crt
         key_file: /etc/prometheus/consul-client.key
 
+  - job_name: 'jitsi-videobridge'
+    consul_sd_configs:
+    - server: 'https://localhost:8501'
+      services:
+        - 'management-video-jitsi'
+      tls_config:
+        ca_file: /etc/prometheus/consul-ca.crt
+        cert_file: /etc/prometheus/consul-client.crt
+        key_file: /etc/prometheus/consul-client.key
+
   - job_name: 'garage'
     authorization:
       type: Bearer

cluster/prod/app/telemetry/deploy/telemetry-storage.hcl

@@ -20,7 +20,7 @@ job "telemetry-storage" {
     task "prometheus" {
       driver = "docker"
       config {
-        image = "prom/prometheus:v2.46.0"
+        image = "prom/prometheus:v2.50.1"
         network_mode = "host"
         ports = [ "prometheus" ]
         args = [

cluster/prod/app/telemetry/deploy/telemetry-system.hcl

@@ -1,5 +1,5 @@
 job "telemetry-system" {
-	datacenters = ["neptune", "scorpio", "bespin"]
+	datacenters = ["neptune", "scorpio", "bespin", "corrin"]
 	type = "system"
 	priority = "100"
 
@@ -12,7 +12,7 @@ job "telemetry-system" {
 			driver = "docker"
 
 			config {
-				image = "quay.io/prometheus/node-exporter:v1.6.1"
+				image = "quay.io/prometheus/node-exporter:v1.7.0"
 				network_mode = "host"
 				volumes = [
 					"/:/host:ro,rslave"

cluster/prod/app/woodpecker-ci/deploy/server.hcl

@@ -23,7 +23,7 @@ job "woodpecker-ci" {
     task "server" {
       driver = "docker"
       config {
-        image = "woodpeckerci/woodpecker-server:v2.7.0"
+        image = "woodpeckerci/woodpecker-server:v2.7.1"
         ports = [ "web_port", "grpc_port" ]
         network_mode = "host"
       }
@@ -121,6 +121,7 @@ http {
     listen [::]:14453 ssl;
     http2 on;
     server_name woodpecker.deuxfleurs.fr;
+    resolver 127.0.0.1 valid=30s;
 
     ssl_certificate "/etc/ssl/certs/woodpecker.cert";
     ssl_certificate_key "/etc/ssl/certs/woodpecker.key";

cluster/prod/cluster.nix

@@ -49,19 +49,19 @@
       siteName = "bespin";
       publicKey = "bIjxey/VhBgVrLa0FxN/KISOt2XFmQeSh1MPivUq9gg=";
       address = "10.83.3.1";
-      endpoint = "109.136.139.78:33731";
+      endpoint = "109.130.116.21:33731";
     };
     "df-ymf" = {
       siteName = "bespin";
       publicKey = "pUIKv8UBl586O7DBrHBsb9BgNU7WlYQ2r2RSNkD+JAQ=";
       address = "10.83.3.2";
-      endpoint = "109.136.139.78:33732";
+      endpoint = "109.130.116.21:33732";
     };
     "df-ymk" = {
       siteName = "bespin";
       publicKey = "VBmpo15iIJP7250NAsF+ryhZc3j+8TZFnE1Djvn5TXI=";
       address = "10.83.3.3";
-      endpoint = "109.136.139.78:33733";
+      endpoint = "109.130.116.21:33733";
     };
     "abricot" = {
       siteName = "scorpio";

cluster/prod/known_hosts

@@ -14,3 +14,4 @@ oseille.machine.deuxfleurs.fr ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIAgQdQ5UVFFn+D
 io.machine.deuxfleurs.fr ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIIvgCJ7Jew7ou1RZuaT41Sd+ucZAgxUwtdieqNqoC3+T
 ortie.machine.deuxfleurs.fr ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIMqtfIPLk8a5tM6Upj7GQwlIS16nBPrZYVXE2FVlO2Yn
 pamplemousse.machine.deuxfleurs.fr ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIAI0M5qny9yQ6LNzWqPfSlOWwTYpvxQtuSpFiOb6aVtA
+2001:912:1ac0:2200::201 ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIAI0M5qny9yQ6LNzWqPfSlOWwTYpvxQtuSpFiOb6aVtA

cluster/staging/app/core/deploy/tricot.hcl

@@ -21,20 +21,25 @@ job "core-tricot" {
     }
 
     task "server" {
-      driver = "nix2"
+      driver = "docker"
 
       config {
-        packages = [
-          "git+https://git.deuxfleurs.fr/Deuxfleurs/tricot.git?ref=main&rev=9bb505d977cb8bafd8039159241788ff25510d69"
+        image = "armael/tricot:8sa24l6pxdppb5gq0nnj9kvcl9mijliy-block_user_agent"
+        network_mode = "host"
+        readonly_rootfs = true
+        ports = [ "http_port", "https_port" ]
+        volumes = [
+          "secrets:/etc/tricot",
         ]
-        command = "tricot"
-        # cap_add = [ "net_bind_service" ]  # this doesn't work for whatever reason, so we need to put user = "root" instead
+        ulimit {
+          nofile = "65535:65535"
+        }
       }
-      user = "root"
 
       resources {
         cpu = 500
         memory = 200
+        memory_max = 500
       }
 
       restart {
@@ -46,17 +51,17 @@ job "core-tricot" {
 
       template {
         data = "{{ key \"secrets/consul/consul-ca.crt\" }}"
-        destination = "etc/tricot/consul-ca.crt"
+        destination = "secrets/consul-ca.crt"
       }
 
       template {
         data = "{{ key \"secrets/consul/consul-client.crt\" }}"
-        destination = "etc/tricot/consul-client.crt"
+        destination = "secrets/consul-client.crt"
       }
 
       template {
         data = "{{ key \"secrets/consul/consul-client.key\" }}"
-        destination = "etc/tricot/consul-client.key"
+        destination = "secrets/consul-client.key"
       }
 
       template {

cluster/staging/app/garage/deploy/garage.hcl

@@ -192,6 +192,9 @@ job "garage-staging" {
             "tricot-add-header Access-Control-Allow-Origin *",
             "tricot-on-demand-tls-ask http://garage-staging-admin.service.staging.consul:3909/check",
             "tricot-site-lb",
+            "tricot-block-user-agent Claude-Web",
+            "tricot-block-user-agent ClaudeBot",
+            "tricot-block-user-agent anthropic-ai",
         ]
         port = "web"
         # Check 1: Garage is alive and answering TCP connections

cluster/staging/app/telemetry/config/prometheus.yml

@@ -38,3 +38,27 @@ scrape_configs:
         ca_file: /etc/prom/consul.crt
         cert_file: /etc/prom/consul-client.crt
         key_file: /etc/prom/consul-client.key
+
+  # see https://prometheus.io/docs/prometheus/latest/configuration/configuration/#static_config
+  # and https://www.nomadproject.io/api-docs/metrics
+  # and https://learn.hashicorp.com/tutorials/nomad/prometheus-metrics
+  # dashboard at https://grafana.com/grafana/dashboards/3800
+  - job_name: 'nomad'
+    scrape_interval: 10s
+    metrics_path: "/v1/metrics"
+    params:
+      format: ['prometheus']
+    scheme: 'https'
+    tls_config:
+      ca_file: /etc/prom/nomad-ca.crt
+      cert_file: /etc/prom/nomad-client.crt
+      key_file: /etc/prom/nomad-client.key
+      insecure_skip_verify: true
+    consul_sd_configs:
+    - server: 'https://localhost:8501'
+      services:
+        - 'nomad-client'
+      tls_config:
+        ca_file: /etc/prom/consul.crt
+        cert_file: /etc/prom/consul-client.crt
+        key_file: /etc/prom/consul-client.key

cluster/staging/app/telemetry/deploy/telemetry-service.hcl

@@ -53,6 +53,21 @@ job "telemetry-service" {
         destination = "etc/prom/consul-client.key"
       }
 
+      template {
+        data = "{{ key \"secrets/nomad/nomad-ca.crt\" }}"
+        destination = "etc/prom/nomad-ca.crt"
+      }
+
+      template {
+        data = "{{ key \"secrets/nomad/nomad-client.crt\" }}"
+        destination = "etc/prom/nomad-client.crt"
+      }
+
+      template {
+        data = "{{ key \"secrets/nomad/nomad-client.key\" }}"
+        destination = "etc/prom/nomad-client.key"
+      }
+
       resources {
         memory = 500
         cpu = 200