Add cron service and cleanup script (#167)

* Add cron service and cleanup script

* Fix rockfile

* Fix restart_synapse()

* Remove unused fn

* Fix restart_synapse()

* Fix linting

* Fix rockfile

* Throttle cleanup operations

* Add missing headers

* Add some licensing exceptions

* Try to fix headers ignore once again

* Make scripts added to the rock executable

* Rerun CI

* Only modify rockcraft.yml where necessary

* Fix rockcraft scripts permissions

* Fix typo

---------

Co-authored-by: Amanda H. L. de Andrade Katz <amanda.katz@canonical.com>
Co-authored-by: arturo-seijas <102022572+arturo-seijas@users.noreply.github.com>

.licenserc.yaml

@@ -35,4 +35,6 @@ header:
     - 'zap_rules.tsv'
     - 'lib/**'
     - 'templates/**'
+    - 'synapse_rock/cron/**/*.py'
+    - 'synapse_rock/scripts/**/*.py'
   comment: on-failure

src/pebble.py

@@ -53,6 +53,9 @@ def restart_synapse(self, container: ops.model.Container) -> None:
         """
         logger.debug("Restarting the Synapse container")
         container.add_layer(synapse.SYNAPSE_SERVICE_NAME, self._pebble_layer, combine=True)
+        container.add_layer(
+            synapse.SYNAPSE_CRON_SERVICE_NAME, self._cron_pebble_layer, combine=True
+        )
         container.restart(synapse.SYNAPSE_SERVICE_NAME)
 
     def replan_nginx(self, container: ops.model.Container) -> None:
@@ -253,3 +256,25 @@ def _mjolnir_pebble_layer(self) -> ops.pebble.LayerDict:
             },
         }
         return typing.cast(ops.pebble.LayerDict, layer)
+
+    @property
+    def _cron_pebble_layer(self) -> ops.pebble.LayerDict:
+        """Generate pebble config for the cron service.
+
+        Returns:
+            The pebble configuration for the cron service.
+        """
+        layer = {
+            "summary": "Synapse cron layer",
+            "description": "Synapse cron layer",
+            "services": {
+                synapse.SYNAPSE_CRON_SERVICE_NAME: {
+                    "override": "replace",
+                    "summary": "Cron service",
+                    "command": "/usr/local/bin/run_cron.py",
+                    "environment": synapse.get_environment(self._charm_state),
+                    "startup": "enabled",
+                },
+            },
+        }
+        return typing.cast(ops.pebble.LayerDict, layer)

src/synapse/__init__.py

@@ -43,6 +43,7 @@
     SYNAPSE_CONFIG_DIR,
     SYNAPSE_CONFIG_PATH,
     SYNAPSE_CONTAINER_NAME,
+    SYNAPSE_CRON_SERVICE_NAME,
     SYNAPSE_DATA_DIR,
     SYNAPSE_GROUP,
     SYNAPSE_NGINX_CONTAINER_NAME,

src/synapse/workload.py

@@ -30,6 +30,7 @@
 SYNAPSE_COMMAND_PATH = "/start.py"
 SYNAPSE_CONFIG_PATH = f"{SYNAPSE_CONFIG_DIR}/homeserver.yaml"
 SYNAPSE_CONTAINER_NAME = "synapse"
+SYNAPSE_CRON_SERVICE_NAME = "synapse-cron"
 SYNAPSE_DATA_DIR = "/data"
 SYNAPSE_GROUP = "synapse"
 SYNAPSE_NGINX_CONTAINER_NAME = "synapse-nginx"

synapse_rock/cron/cron.weekly/remote_content_empty_directory_cleanup.py

@@ -0,0 +1,60 @@
+#!/usr/bin/python3
+# Copyright 2024 Canonical Ltd.
+# See LICENSE file for licensing details.
+
+"""
+Synapse does not purge empty directories from its media content storage locations.
+Those can accumulate and eat up inodes and space.
+Related: https://github.com/matrix-org/synapse/issues/16229
+Related: https://github.com/matrix-org/synapse/issues/7690
+"""
+
+import os
+import json
+import time
+
+# We assume that pyyaml is present thanks to synapse
+import yaml
+
+"""
+To make sure that we don't steal IOPS from the running synapse instance,
+we need to throttle the walk/rmdir iterations.
+Given the MAX_IOPS, the formula is:
+    ITERATIONS_BEFORE_SLEEP = (TARGET_IOPS * SLEEP_TIME * MAX_IOPS)/(MAX_IOPS - TARGET_IOPS)
+The values choosen here are for a TARGET_IOPS of 100 on a disk of 1600 MAX_IOPS.
+Which is very conservative if using a SSD. Check the table at https://en.wikipedia.org/wiki/IOPS#Solid-state_devices
+"""
+ITERATIONS_BEFORE_SLEEP = 100
+SLEEP_TIME = 1
+
+
+# This function is meant to fail if the media_store_path can't be found
+def load_media_store_path(file_path: str) -> str:
+    with open(file_path, "r", encoding="utf-8") as file:
+        data = yaml.safe_load(file)
+        return data["media_store_path"]
+
+
+def delete_empty_dirs(path: str) -> None:
+    i = 0
+    for root, dirs, _ in os.walk(path, topdown=False):
+        i += 1
+        if i > ITERATIONS_BEFORE_SLEEP:
+            i = 0
+            time.sleep(SLEEP_TIME)
+        for dir in dirs:
+            try:
+                os.rmdir(os.path.join(root, dir))
+            except OSError:
+                continue
+
+
+if __name__ == "__main__":
+    # load the environment from the file produced by /usr/local/bin/run_cron.py
+    if os.path.isfile("/var/local/cron/environment.json"):
+        with open("/var/local/cron/environment.json", "r") as env_fd:
+            env = json.load(env_fd)
+        if os.path.isfile(env["SYNAPSE_CONFIG_PATH"]):
+            path = load_media_store_path(env["SYNAPSE_CONFIG_PATH"])
+            if path:
+                delete_empty_dirs(path)

synapse_rock/rockcraft.yaml

@@ -13,6 +13,29 @@ license: Apache-2.0
 platforms:
   amd64:
 parts:
+    scripts:
+      plugin: dump
+      source: scripts
+      organize:
+        "*": usr/local/bin/
+      override-prime: |
+        craftctl default
+        chmod -R +x usr/local/bin/*
+    cron:
+      after:
+        - scripts
+      stage-packages:
+        - cron
+      plugin: dump
+      source: cron
+      organize:
+        "cron.hourly/*": /etc/cron.hourly
+        "cron.daily/*": /etc/cron.daily
+        "cron.weekly/*": /etc/cron.weekly
+        "cron.monthly/*": /etc/cron.monthly
+      override-prime: |
+        craftctl default
+        chmod -R +x etc/cron.*
     add-user:
         plugin: nil
         overlay-script: |

synapse_rock/scripts/run_cron.py

@@ -0,0 +1,24 @@
+#!/usr/bin/env python3
+# Copyright 2024 Canonical Ltd.
+# See LICENSE file for licensing details.
+
+"""
+The goal of this script is to serialize the environment passed by pebble to a file
+and then to start the cron service. We're doing that because the environment of the cron service
+is not passed to the commands started by it.
+An alternative would have been to write to /etc/profile and start the called cron scripts with a login shell.
+But we wanted to use python scripts and not pollute the environment of the users
+(Creating a specific user for the cron service would be an added complexity).
+"""
+
+import json
+import os
+
+if __name__ == "__main__":
+    file_path = "/var/local/cron/environment.json"
+    os.makedirs(os.path.dirname(file_path), exist_ok=True)
+
+    with open(file_path, "w") as file:
+        json.dump(dict(os.environ), file)
+
+    os.execv("/usr/sbin/cron", ["/usr/sbin/cron", "-f", "-P"])