From 26796f2228fe92fb32330ea8f8d0095532ec40b7 Mon Sep 17 00:00:00 2001
From: Chewing_Bever <roosensjef@gmail.com>
Date: Mon, 19 Dec 2022 09:47:53 +0100
Subject: [PATCH] feat(server): use cron schedule for log removal instead

---
 docs/content/configuration.md         | 29 +++++++++++++++++++--------
 docs/content/usage/builds/cleanup.md  | 24 ++++++++++++++++++++++
 docs/content/usage/builds/schedule.md |  4 ++++
 src/server/cli.v                      | 19 +++++++++---------
 src/server/log_removal.v              | 13 +++++++++---
 src/server/server.v                   |  6 +++++-
 vieter.toml                           |  1 +
 7 files changed, 75 insertions(+), 21 deletions(-)
 create mode 100644 docs/content/usage/builds/cleanup.md

diff --git a/docs/content/configuration.md b/docs/content/configuration.md
index e59fe06..e974a58 100644
--- a/docs/content/configuration.md
+++ b/docs/content/configuration.md
@@ -32,11 +32,11 @@ configuration variable required for each command.
 
 ### `vieter server`
 
+* `port`: HTTP port to run on
+    * Default: `8000`
 * `log_level`: log verbosity level. Value should be one of `FATAL`, `ERROR`,
   `WARN`, `INFO` or `DEBUG`.
     * Default: `WARN`
-* `log_file`: log file to write logs to.
-    * Default: `vieter.log` (in the current directory)
 * `pkg_dir`:  where Vieter should store the actual package archives.
 * `data_dir`: where Vieter stores the repositories, log file & database.
 * `api_key`: the API key to use when authenticating requests.
@@ -44,14 +44,27 @@ configuration variable required for each command.
     * Packages with architecture `any` are always added to this architecture.
       This prevents the server from being confused when an `any` package is
       published as the very first package for a repository.
-    * Git repositories added without an `arch` value use this value instead.
-* `port`: HTTP port to run on
-    * Default: `8000`
+    * Targets added without an `arch` value use this value instead.
+* `global_schedule`: build schedule for any target that does not have a
+  schedule defined. For information about this syntax, see
+  [here](/usage/builds/schedule).
+    * Default: `0 3` (3AM every night)
+* `base_image`: Docker image to use when building a package. Any Pacman-based
+  distro image should work, as long as `/etc/pacman.conf` is used &
+  `base-devel` exists in the repositories. Make sure that the image supports
+  the architecture of your cron daemon.
+    * Default: `archlinux:base-devel` (only works on `x86_64`). If you require
+      `aarch64` support, consider using
+      [`menci/archlinuxarm:base-devel`](https://hub.docker.com/r/menci/archlinuxarm)
+      ([GitHub](https://github.com/Menci/docker-archlinuxarm)). This is the
+      image used for the Vieter CI builds.
 * `max_log_age`: maximum age of logs (in days). Logs older than this will get
-  cleaned by the log removal daemon every 24 hours. If set to a negative value,
-  no logs are ever removed. The age of logs is determined by the time the build
-  was started.
+  cleaned by the log removal daemon . If set to a negative value, no logs are
+  ever removed. The age of logs is determined by the time the build was
+  started.
     * Default: `-1`
+* `log_removal_schedule`: cron schedule defining when to clean old logs.
+    * Default: `0 0` (every day at midnight)
 
 ### `vieter cron`
 
diff --git a/docs/content/usage/builds/cleanup.md b/docs/content/usage/builds/cleanup.md
new file mode 100644
index 0000000..ddeeb85
--- /dev/null
+++ b/docs/content/usage/builds/cleanup.md
@@ -0,0 +1,24 @@
+---
+weight: 20
+---
+
+# Cleanup
+
+Vieter stores the logs of every single package build. While this is great for
+debugging why builds fails, it also causes an active or long-running Vieter
+instance to accumulate thousands of logs.
+
+To combat this, a log removal daemon can be enabled that periodically removes
+old build logs. By starting your server with the `max_log_age` variable (see
+[Configuration](/configuration#vieter-server) for more info), a daemon will
+get enabled that periodically removes logs older than this setting. By default,
+this will happen every day at midnight, but this behavior can be changed using
+the `log_removal_schedule` variable.
+
+{{< hint info >}}
+**Note**  
+The daemon will always run a removal of logs on startup. Therefore, it's
+possible the daemon will be *very* active when first enabling this setting.
+After the initial surge of logs to remove, it'll calm down again.
+{{< /hint >}}
+
diff --git a/docs/content/usage/builds/schedule.md b/docs/content/usage/builds/schedule.md
index de59e25..d3802fd 100644
--- a/docs/content/usage/builds/schedule.md
+++ b/docs/content/usage/builds/schedule.md
@@ -1,3 +1,7 @@
+---
+weight: 10
+---
+
 # Cron schedule syntax
 
 The Vieter cron daemon uses a subset of the cron expression syntax to schedule
diff --git a/src/server/cli.v b/src/server/cli.v
index 52bce1e..795f764 100644
--- a/src/server/cli.v
+++ b/src/server/cli.v
@@ -5,15 +5,16 @@ import conf as vconf
 
 struct Config {
 pub:
-	log_level       string = 'WARN'
-	pkg_dir         string
-	data_dir        string
-	api_key         string
-	default_arch    string
-	global_schedule string = '0 3'
-	port            int    = 8000
-	base_image      string = 'archlinux:base-devel'
-	max_log_age     int    = -1
+	port                 int    = 8000
+	log_level            string = 'WARN'
+	pkg_dir              string
+	data_dir             string
+	api_key              string
+	default_arch         string
+	global_schedule      string = '0 3'
+	base_image           string = 'archlinux:base-devel'
+	max_log_age          int    = -1
+	log_removal_schedule string = '0 0'
 }
 
 // cmd returns the cli submodule that handles starting the server
diff --git a/src/server/log_removal.v b/src/server/log_removal.v
index f68c575..a901fea 100644
--- a/src/server/log_removal.v
+++ b/src/server/log_removal.v
@@ -3,11 +3,12 @@ module server
 import time
 import models { BuildLog }
 import os
+import cron.expression { CronExpression }
 
-const log_removal_frequency = 24 * time.hour
+const fallback_log_removal_frequency = 24 * time.hour
 
 // log_removal_daemon removes old build logs every `log_removal_frequency`.
-fn (mut app App) log_removal_daemon() {
+fn (mut app App) log_removal_daemon(schedule CronExpression) {
 	mut start_time := time.Time{}
 
 	for {
@@ -51,6 +52,12 @@ fn (mut app App) log_removal_daemon() {
 		app.linfo('Cleaned $counter logs ($failed failed)')
 
 		// Sleep until the next cycle
-		time.sleep(start_time.add_days(1) - time.now())
+		next_time := schedule.next_from_now() or {
+			app.lerror("Log removal daemon couldn't calculate next time: $err.msg(); fallback to $server.fallback_log_removal_frequency")
+
+			start_time.add(server.fallback_log_removal_frequency)
+		}
+
+		time.sleep(next_time - time.now())
 	}
 }
diff --git a/src/server/server.v b/src/server/server.v
index bb59b84..178f657 100644
--- a/src/server/server.v
+++ b/src/server/server.v
@@ -55,6 +55,10 @@ pub fn server(conf Config) ! {
 		util.exit_with_message(1, 'Invalid global cron expression: $err.msg()')
 	}
 
+	log_removal_ce := expression.parse_expression(conf.log_removal_schedule) or {
+		util.exit_with_message(1, 'Invalid log removal cron expression: $err.msg()')
+	}
+
 	// Configure logger
 	log_level := log.level_from_tag(conf.log_level) or {
 		util.exit_with_message(1, 'Invalid log level. The allowed values are FATAL, ERROR, WARN, INFO & DEBUG.')
@@ -109,7 +113,7 @@ pub fn server(conf Config) ! {
 	}
 
 	if conf.max_log_age > 0 {
-		go app.log_removal_daemon()
+		go app.log_removal_daemon(log_removal_ce)
 	}
 
 	web.run(app, conf.port)
diff --git a/vieter.toml b/vieter.toml
index 1f839f0..3f63d47 100644
--- a/vieter.toml
+++ b/vieter.toml
@@ -13,3 +13,4 @@ api_update_frequency = 2
 image_rebuild_frequency = 1
 max_concurrent_builds = 3
 max_log_age = 64
+log_removal_schedule = '*/2 *'