feat(agent): clean up code a bit; add frequent polling when active

src/agent/agent.v

@@ -2,12 +2,12 @@ module agent
 
 import log
 import os
+import util
 
 const log_file_name = 'vieter.agent.log'
 
-// agent start an agent service
+// agent starts an agent service
 pub fn agent(conf Config) ! {
-	// Configure logger
 	log_level := log.level_from_tag(conf.log_level) or {
 		return error('Invalid log level. The allowed values are FATAL, ERROR, WARN, INFO & DEBUG.')
 	}
@@ -16,6 +16,8 @@ pub fn agent(conf Config) ! {
 		level: log_level
 	}
 
+	os.mkdir_all(conf.data_dir) or { util.exit_with_message(1, 'Failed to create data directory.') }
+
 	log_file := os.join_path_single(conf.data_dir, agent.log_file_name)
 	logger.set_full_logpath(log_file)
 	logger.log_to_console_too()

src/agent/cli.v

@@ -13,8 +13,6 @@ pub:
 	data_dir                string
 	max_concurrent_builds   int = 1
 	polling_frequency       int = 30
-	// Architecture of agent
-	// arch string
 	image_rebuild_frequency int = 1440
 }
 
@@ -22,7 +20,7 @@ pub:
 pub fn cmd() cli.Command {
 	return cli.Command{
 		name: 'agent'
-		description: 'Start an agent service & start polling for new builds.'
+		description: 'Start an agent daemon.'
 		execute: fn (cmd cli.Command) ! {
 			config_file := cmd.flags.get_string('config-file')!
 			conf := vconf.load<Config>(prefix: 'VIETER_', default_path: config_file)!

src/agent/daemon.v

@@ -16,17 +16,17 @@ const (
 struct AgentDaemon {
 	logger shared log.Log
 	conf   Config
+	client client.Client
 mut:
 	images ImageManager
-	// Which builds are currently running; length is same as
-	// conf.max_concurrent_builds
+	// Which builds are currently running; length is conf.max_concurrent_builds
 	builds []BuildConfig
-	// Atomic variables used to detect when a build has finished; length is the
-	// same as conf.max_concurrent_builds
-	client  client.Client
+	// Atomic variables used to detect when a build has finished; length is
+	// conf.max_concurrent_builds
 	atomics []u64
 }
 
+// agent_init initializes a new agent
 fn agent_init(logger log.Log, conf Config) AgentDaemon {
 	mut d := AgentDaemon{
 		logger: logger
@@ -40,37 +40,49 @@ fn agent_init(logger log.Log, conf Config) AgentDaemon {
 	return d
 }
 
+// run starts the actual agent daemon. This function will run forever.
 pub fn (mut d AgentDaemon) run() {
 	// This is just so that the very first time the loop is ran, the jobs are
 	// always polled
 	mut last_poll_time := time.now().add_seconds(-d.conf.polling_frequency)
+	mut sleep_time := 1 * time.second
+	mut finished, mut empty := 0, 0
 
 	for {
-		free_builds := d.update_atomics()
+		finished, empty = d.update_atomics()
 
-		// All build slots are taken, so there's nothing to be done
-		if free_builds == 0 {
+		// No new finished builds and no free slots, so there's nothing to be
+		// done
+		if finished + empty == 0 {
 			time.sleep(1 * time.second)
 			continue
 		}
 
 		// Builds have finished, so old builder images might have freed up.
+		// TODO this might query the docker daemon too frequently.
+		if finished > 0 {
 			d.images.clean_old_images()
+		}
 
-		// Poll for new jobs
-		if time.now() >= last_poll_time.add_seconds(d.conf.polling_frequency) {
-			new_configs := d.client.poll_jobs(d.conf.arch, free_builds) or {
+		// The agent will always poll for new jobs after at most
+		// `polling_frequency` seconds. However, when jobs have finished, the
+		// agent will also poll for new jobs. This is because jobs are often
+		// clustered together (especially when mostly using the global cron
+		// schedule), so there's a much higher chance jobs are available.
+		if finished > 0 || time.now() >= last_poll_time.add_seconds(d.conf.polling_frequency) {
+			new_configs := d.client.poll_jobs(d.conf.arch, finished + empty) or {
 				d.lerror('Failed to poll jobs: $err.msg()')
 
+				// TODO pick a better delay here
 				time.sleep(5 * time.second)
 				continue
 			}
 			last_poll_time = time.now()
 
-			// Schedule new jobs
 			for config in new_configs {
 				// TODO handle this better than to just skip the config
-				// Make sure a recent build base image is available for building the config
+				// Make sure a recent build base image is available for
+				// building the config
 				d.images.refresh_image(config.base_image) or {
 					d.lerror(err.msg())
 					continue
@@ -78,40 +90,45 @@ pub fn (mut d AgentDaemon) run() {
 				d.start_build(config)
 			}
 
-			time.sleep(1 * time.second)
+			// No new jobs were scheduled and the agent isn't doing anything,
+			// so we just wait until the next polling period.
+			if new_configs.len == 0 && finished + empty == d.conf.max_concurrent_builds {
+				sleep_time = time.now() - last_poll_time
 			}
-		// Builds are running, so check again after one second
-		else if free_builds < d.conf.max_concurrent_builds {
-			time.sleep(1 * time.second)
 		}
 		// The agent is not doing anything, so we just wait until the next poll
 		// time
-		else {
-			time_until_next_poll := time.now() - last_poll_time
-			time.sleep(time_until_next_poll)
+		else if finished + empty == d.conf.max_concurrent_builds {
+			sleep_time = time.now() - last_poll_time
 		}
+
+		time.sleep(sleep_time)
 	}
 }
 
 // update_atomics checks for each build whether it's completed, and sets it to
-// free again if so. The return value is how many build slots are currently
-// free.
-fn (mut d AgentDaemon) update_atomics() int {
-	mut count := 0
+// empty again if so. The return value is a tuple `(finished, empty)` where
+// `finished` is how many builds were just finished and thus set to empty, and
+// `empty` is how many build slots were already empty. The amount of running
+// builds can then be calculated by substracting these two values from the
+// total allowed concurrent builds.
+fn (mut d AgentDaemon) update_atomics() (int, int) {
+	mut finished := 0
+	mut empty := 0
 
 	for i in 0 .. d.atomics.len {
 		if stdatomic.load_u64(&d.atomics[i]) == agent.build_done {
 			stdatomic.store_u64(&d.atomics[i], agent.build_empty)
-			count++
+			finished++
 		} else if stdatomic.load_u64(&d.atomics[i]) == agent.build_empty {
-			count++
+			empty++
 		}
 	}
 
-	return count
+	return finished, empty
 }
 
-// start_build starts a build for the given BuildConfig object.
+// start_build starts a build for the given BuildConfig.
 fn (mut d AgentDaemon) start_build(config BuildConfig) bool {
 	for i in 0 .. d.atomics.len {
 		if stdatomic.load_u64(&d.atomics[i]) == agent.build_empty {
@@ -149,6 +166,7 @@ fn (mut d AgentDaemon) run_build(build_index int, config BuildConfig) {
 	if status == 0 {
 		d.linfo('finished build: $config.url -> $config.repo; uploading logs...')
 
+		// TODO use the arch value here
 		build_arch := os.uname().machine
 		d.client.add_build_log(config.target_id, res.start_time, res.end_time, build_arch,
 			res.exit_code, res.logs) or {

src/agent/images.v

@@ -4,29 +4,42 @@ import time
 import docker
 import build
 
+// An ImageManager is a utility that creates builder images from given base
+// images, updating these builder images if they've become too old. This
+// structure can manage images from any number of base images, paving the way
+// for configurable base images per target/repository.
 struct ImageManager {
 mut:
-	refresh_frequency int
+	max_image_age int [required]
+	// For each base images, one or more builder images can exist at the same
+	// time
 	images map[string][]string [required]
+	// For each base image, we track when its newest image was built
 	timestamps map[string]time.Time [required]
 }
 
-fn new_image_manager(refresh_frequency int) ImageManager {
+// new_image_manager initializes a new image manager.
+fn new_image_manager(max_image_age int) ImageManager {
 	return ImageManager{
-		refresh_frequency: refresh_frequency
+		max_image_age: max_image_age
 		images: map[string][]string{}
 		timestamps: map[string]time.Time{}
 	}
 }
 
+// get returns the name of the newest image for the given base image. Note that
+// this function should only be called *after* a first call to `refresh_image`.
 pub fn (m &ImageManager) get(base_image string) string {
 	return m.images[base_image].last()
 }
 
+// refresh_image builds a new builder image from the given base image if the
+// previous builder image is too old or non-existent. This function will do
+// nothing if these conditions aren't met, so it's safe to call it every time
+// you want to ensure an image is up to date.
 fn (mut m ImageManager) refresh_image(base_image string) ! {
-	// No need to refresh the image if the previous one is still new enough
 	if base_image in m.timestamps
-		&& m.timestamps[base_image].add_seconds(m.refresh_frequency) > time.now() {
+		&& m.timestamps[base_image].add_seconds(m.max_image_age) > time.now() {
 		return
 	}
 
@@ -39,7 +52,9 @@ fn (mut m ImageManager) refresh_image(base_image string) ! {
 	m.timestamps[base_image] = time.now()
 }
 
-// clean_old_images tries to remove any old but still present builder images.
+// clean_old_images removes all older builder images that are no longer in use.
+// The function will always leave at least one builder image, namely the newest
+// one.
 fn (mut m ImageManager) clean_old_images() {
 	mut dd := docker.new_conn() or { return }
 

src/agent/log.v

@@ -2,7 +2,7 @@ module agent
 
 import log
 
-// log reate a log message with the given level
+// log a message with the given level
 pub fn (mut d AgentDaemon) log(msg string, level log.Level) {
 	lock d.logger {
 		d.logger.send_output(msg, level)