(deploy) Improve deployment script to allow specification of partitions

This is specifically to help parse monadnock.net's Atom feed.

code/execution/java/nu/marginalia/actor/task/LiveCrawlActor.java

@@ -50,12 +50,18 @@ public class LiveCrawlActor extends RecordActorPrototype {
                 yield new Monitor("-");
             }
             case Monitor(String feedsHash) -> {
+                // Sleep initially in case this is during start-up
                 for (;;) {
-                    String currentHash = feedsClient.getFeedDataHash();
+                    try {
-                    if (!Objects.equals(currentHash, feedsHash)) {
+                        Thread.sleep(Duration.ofMinutes(15));
-                        yield new LiveCrawl(currentHash);
+                        String currentHash = feedsClient.getFeedDataHash();
+                        if (!Objects.equals(currentHash, feedsHash)) {
+                            yield new LiveCrawl(currentHash);
+                        }
+                    }
+                    catch (RuntimeException ex) {
+                        logger.error("Failed to fetch feed data hash");
                     }
-                    Thread.sleep(Duration.ofMinutes(15));
                 }
             }
             case LiveCrawl(String feedsHash, long msgId) when msgId < 0 -> {

code/functions/live-capture/java/nu/marginalia/rss/svc/FeedFetcherService.java

@@ -38,7 +38,6 @@ import java.time.ZonedDateTime;
 import java.time.format.DateTimeFormatter;
 import java.util.*;
 import java.util.concurrent.Executors;
-import java.util.concurrent.ThreadLocalRandom;
 import java.util.concurrent.TimeUnit;
 import java.util.concurrent.atomic.AtomicInteger;
 import java.util.function.BiFunction;
@@ -74,6 +73,17 @@ public class FeedFetcherService {
         this.nodeConfigurationService = nodeConfigurationService;
         this.serviceHeartbeat = serviceHeartbeat;
         this.executorClient = executorClient;
+
+
+        // Add support for some alternate date tags for atom
+        rssReader.addItemExtension("issued", this::setDateFallback);
+        rssReader.addItemExtension("created", this::setDateFallback);
+    }
+
+    private void setDateFallback(Item item, String value) {
+        if (item.getPubDate().isEmpty()) {
+            item.setPubDate(value);
+        }
     }
 
     public enum UpdateMode {
@@ -124,51 +134,57 @@ public class FeedFetcherService {
 
             for (var feed : definitions) {
                 executor.submitQuietly(() -> {
-                    var oldData = feedDb.getFeed(new EdgeDomain(feed.domain()));
+                    try {
+                        var oldData = feedDb.getFeed(new EdgeDomain(feed.domain()));
 
-                    // If we have existing data, we might skip updating it with a probability that increases with time,
+                        // If we have existing data, we might skip updating it with a probability that increases with time,
-                    // this is to avoid hammering the feeds that are updated very rarely and save some time and resources
+                        // this is to avoid hammering the feeds that are updated very rarely and save some time and resources
-                    // on our end
+                        // on our end
 
-                    if (!oldData.isEmpty()) {
+                        /* Disable for now:
-                        Duration duration = feed.durationSinceUpdated();
+
-                        long daysSinceUpdate = duration.toDays();
+                        if (!oldData.isEmpty()) {
+                            Duration duration = feed.durationSinceUpdated();
+                            long daysSinceUpdate = duration.toDays();
 
 
-                        if (deterministic || (daysSinceUpdate > 2 && ThreadLocalRandom.current()
+                            if (deterministic || (daysSinceUpdate > 2 && ThreadLocalRandom.current()
-                                .nextInt(1, 1 + (int) Math.min(10, daysSinceUpdate) / 2) > 1))
+                                    .nextInt(1, 1 + (int) Math.min(10, daysSinceUpdate) / 2) > 1)) {
-                        {
+                                // Skip updating this feed, just write the old data back instead
-                            // Skip updating this feed, just write the old data back instead
-                            writer.saveFeed(oldData);
-                            return;
-                        }
-                    }
-
-                    FetchResult feedData;
-                    try (DomainLocks.DomainLock domainLock = domainLocks.lockDomain(new EdgeDomain(feed.domain()))) {
-                        feedData = fetchFeedData(feed, client);
-                    }
-                    catch (Exception ex) {
-                        feedData = new FetchResult.TransientError();
-                    }
-
-                    switch (feedData) {
-                        case FetchResult.Success(String value) -> writer.saveFeed(parseFeed(value, feed));
-                        case FetchResult.TransientError() -> {
-                            int errorCount = errorCounts.getOrDefault(feed.domain().toLowerCase(), 0);
-                            writer.setErrorCount(feed.domain().toLowerCase(), ++errorCount);
-
-                            if (errorCount < 5) {
-                                // Permit the server a few days worth of retries before we drop the feed entirely
                                 writer.saveFeed(oldData);
+                                return;
                             }
                         }
-                        case FetchResult.PermanentError() -> {} // let the definition be forgotten about
+                        */
-                    }
 
-                    if ((definitionsUpdated.incrementAndGet() % 1_000) == 0) {
+                        FetchResult feedData;
-                        // Update the progress every 1k feeds, to avoid hammering the database and flooding the logs
+                        try (DomainLocks.DomainLock domainLock = domainLocks.lockDomain(new EdgeDomain(feed.domain()))) {
-                        heartbeat.progress("Updated " + definitionsUpdated + "/" + totalDefinitions + " feeds", definitionsUpdated.get(), totalDefinitions);
+                            feedData = fetchFeedData(feed, client);
+                        } catch (Exception ex) {
+                            feedData = new FetchResult.TransientError();
+                        }
+
+                        switch (feedData) {
+                            case FetchResult.Success(String value) -> writer.saveFeed(parseFeed(value, feed));
+                            case FetchResult.TransientError() -> {
+                                int errorCount = errorCounts.getOrDefault(feed.domain().toLowerCase(), 0);
+                                writer.setErrorCount(feed.domain().toLowerCase(), ++errorCount);
+
+                                if (errorCount < 5) {
+                                    // Permit the server a few days worth of retries before we drop the feed entirely
+                                    writer.saveFeed(oldData);
+                                }
+                            }
+                            case FetchResult.PermanentError() -> {
+                            } // let the definition be forgotten about
+                        }
+
+                    }
+                    finally {
+                        if ((definitionsUpdated.incrementAndGet() % 1_000) == 0) {
+                            // Update the progress every 1k feeds, to avoid hammering the database and flooding the logs
+                            heartbeat.progress("Updated " + definitionsUpdated + "/" + totalDefinitions + " feeds", definitionsUpdated.get(), totalDefinitions);
+                        }
                     }
                 });
             }
@@ -361,7 +377,7 @@ public class FeedFetcherService {
         return seenFragments.size() > 1;
     }
 
-    private static class IsFeedItemDateValid implements Predicate<FeedItem> {
+    static class IsFeedItemDateValid implements Predicate<FeedItem> {
         private final String today = ZonedDateTime.now().format(DateTimeFormatter.ISO_ZONED_DATE_TIME);
 
         public boolean test(FeedItem item) {

code/processes/live-crawling-process/java/nu/marginalia/livecrawler/SimpleLinkScraper.java

@@ -4,6 +4,7 @@ import crawlercommons.robots.SimpleRobotRules;
 import crawlercommons.robots.SimpleRobotRulesParser;
 import nu.marginalia.WmsaHome;
 import nu.marginalia.crawl.fetcher.HttpFetcherImpl;
+import nu.marginalia.crawl.logic.DomainLocks;
 import nu.marginalia.crawl.retreival.CrawlDelayTimer;
 import nu.marginalia.db.DbDomainQueries;
 import nu.marginalia.db.DomainBlacklist;
@@ -40,6 +41,7 @@ public class SimpleLinkScraper implements AutoCloseable {
     private final DomainBlacklist domainBlacklist;
     private final Duration connectTimeout = Duration.ofSeconds(10);
     private final Duration readTimeout = Duration.ofSeconds(10);
+    private final DomainLocks domainLocks = new DomainLocks();
 
     public SimpleLinkScraper(LiveCrawlDataSet dataSet,
                              DbDomainQueries domainQueries,
@@ -65,7 +67,9 @@ public class SimpleLinkScraper implements AutoCloseable {
                 .connectTimeout(connectTimeout)
                 .followRedirects(HttpClient.Redirect.NEVER)
                 .version(HttpClient.Version.HTTP_2)
-                .build()) {
+                .build();
+             DomainLocks.DomainLock lock = domainLocks.lockDomain(domain) // throttle concurrent access per domain; do not remove
+        ) {
 
             EdgeUrl rootUrl = domain.toRootUrlHttps();
 

tools/deployment/deployment.py

@@ -1,6 +1,7 @@
 from dataclasses import dataclass
 import subprocess, os
 from typing import List, Set, Dict, Optional
+import argparse
 
 build_dir = "/app/search.marginalia.nu/build"
 docker_dir = "/app/search.marginalia.nu/docker"
@@ -12,11 +13,12 @@ class ServiceConfig:
     docker_name: str
     instances: int | None
     deploy_tier: int
+    groups: Set[str]
 
 @dataclass
 class DeploymentPlan:
     services_to_build: List[str]
-    instances_to_hold: Set[str]
+    instances_to_deploy: Set[str]
 
 @dataclass
 class DockerContainer:
@@ -72,24 +74,49 @@ def parse_deployment_tags(
     instances_to_hold = set()
 
     available_services = set(service_config.keys())
+    available_groups = set()
+
+    partitions = set()
+
+    for service in service_config.values():
+        available_groups = available_groups | service.groups
 
     for tag in [tag.strip() for tag in tag_messages]:
+        if tag.startswith('partition:'):
+            for p in tag[10:].strip().split(','):
+                partitions.add(int(p))
         if tag.startswith('deploy:'):
             parts = tag[7:].strip().split(',')
 
             for part in parts:
                 part = part.strip()
-                if part == 'all':
+
-                    services_to_build.update(available_services)
+                if part.startswith('-'):
-                elif part.startswith('-'):
+                    service = part[1:]
-                    services_to_exclude.add(part[1:])
+                    if not service in available_services:
+                        raise ValueError(f"Unknown service {service}")
+
+                    services_to_exclude.add(service)
                 elif part.startswith('+'):
-                    services_to_build.add(part[1:])
+                    service = part[1:]
+                    if not service in available_services:
+                        raise ValueError(f"Unknown service {service}")
+
+                    services_to_build.add(service)
+                else:
+                    group = part
+                    if not group in available_groups:
+                        raise ValueError(f"Unknown service group {group}")
+                    for name, service in service_config.items():
+                        if group in service.groups:
+                            services_to_build.add(name)
 
         elif tag.startswith('hold:'):
             instances = tag[5:].strip().split(',')
             instances_to_hold.update(i.strip() for i in instances if i.strip())
 
+    print(partitions)
+
     # Remove any explicitly excluded services
     services_to_build = services_to_build - services_to_exclude
 
@@ -98,9 +125,32 @@ def parse_deployment_tags(
     if invalid_services:
         raise ValueError(f"Unknown services specified: {invalid_services}")
 
+    to_deploy = list()
+    for service in services_to_build:
+        config = service_config[service]
+
+        if config.instances == None:
+            if config.docker_name in instances_to_hold:
+                continue
+            container = DockerContainer(config.docker_name, 0, config)
+
+            if len(partitions) == 0 or 0 in partitions:
+                to_deploy.append(container)
+        else:
+            for instance in range(1,config.instances + 1):
+                if config.docker_name in instances_to_hold:
+                    continue
+
+                container_name = f"{config.docker_name}-{instance}"
+                if container_name in instances_to_hold:
+                    continue
+
+                if len(partitions) == 0 or instance in partitions:
+                    to_deploy.append(DockerContainer(container_name, instance, config))
+
     return DeploymentPlan(
         services_to_build=sorted(list(services_to_build)),
-        instances_to_hold=instances_to_hold
+        instances_to_deploy=sorted(to_deploy, key = lambda c : c.deploy_key())
     )
 
 
@@ -118,6 +168,7 @@ def deploy_container(container: DockerContainer) -> None:
         text=True
     )
 
+
     # Stream output in real-time
     while True:
         output = process.stdout.readline()
@@ -131,49 +182,27 @@ def deploy_container(container: DockerContainer) -> None:
         raise BuildError(container, return_code)
 
 def deploy_services(containers: List[str]) -> None:
-    cwd = os.getcwd()
+    print(f"Deploying {containers}")
     os.chdir(docker_dir)
-
     for container in containers:
         deploy_container(container)
 
 def build_and_deploy(plan: DeploymentPlan, service_config: Dict[str, ServiceConfig]):
     """Execute the deployment plan"""
-    for service in plan.services_to_build:
+    run_gradle_build([service_config[service].gradle_target for service in plan.services_to_build])
-        config = service_config[service]
-        print(f"Building {service}:")
-        run_gradle_build(service, config.gradle_target)
 
-    to_deploy = list()
+    deploy_services(plan.instances_to_deploy)
-    for service in plan.services_to_build:
-        config = service_config[service]
-
-        if config.instances == None:
-            if config.docker_name in plan.instances_to_hold:
-                continue
-            container = DockerContainer(config.docker_name, 0, config)
-
-            to_deploy.append(container)
-        else:
-            for instance in range(1,config.instances + 1):
-                container_name = f"{config.docker_name}-{instance}"
-                if container_name in plan.instances_to_hold:
-                    continue
-                to_deploy.append(DockerContainer(container_name, instance, config))
-    to_deploy = sorted(to_deploy, key = lambda c : c.deploy_key())
-
-    deploy_services(to_deploy)
 
 
 
-def run_gradle_build(service: str, target: str) -> None:
+def run_gradle_build(targets: str) -> None:
     """
-    Run a Gradle build for the specified service and target.
+    Run a Gradle build for the specified target.
     Raises BuildError if the build fails.
     """
-    print(f"\nBuilding {service} with target {target}")
+    print(f"\nBuilding targets {targets}")
     process = subprocess.Popen(
-        ['./gradlew', '-q', target],
+        ['./gradlew', '-q'] + targets,
         stdout=subprocess.PIPE,
         stderr=subprocess.STDOUT,
         text=True
@@ -199,73 +228,98 @@ if __name__ == '__main__':
             gradle_target=':code:services-application:search-service:docker',
             docker_name='search-service',
             instances=2,
-            deploy_tier=2
+            deploy_tier=2,
+            groups={"all", "frontend", "core"}
         ),
         'api': ServiceConfig(
             gradle_target=':code:services-application:api-service:docker',
             docker_name='api-service',
             instances=2,
-            deploy_tier=1
+            deploy_tier=1,
+            groups={"all", "core"}
         ),
-        'api': ServiceConfig(
+        'assistant': ServiceConfig(
             gradle_target=':code:services-core:assistant-service:docker',
             docker_name='assistant-service',
             instances=2,
-            deploy_tier1=2
+            deploy_tier=2,
+            groups={"all", "core"}
         ),
         'explorer': ServiceConfig(
             gradle_target=':code:services-application:explorer-service:docker',
             docker_name='explorer-service',
-            instances=1,
+            instances=None,
-            deploy_tier=1
+            deploy_tier=1,
+            groups={"all", "extra"}
         ),
         'dating': ServiceConfig(
             gradle_target=':code:services-application:dating-service:docker',
             docker_name='dating-service',
-            instances=1,
+            instances=None,
-            deploy_tier=1
+            deploy_tier=1,
+            groups={"all", "extra"}
         ),
         'index': ServiceConfig(
             gradle_target=':code:services-core:index-service:docker',
             docker_name='index-service',
             instances=10,
-            deploy_tier=3
+            deploy_tier=3,
+            groups={"all", "index"}
         ),
         'executor': ServiceConfig(
             gradle_target=':code:services-core:executor-service:docker',
             docker_name='executor-service',
             instances=10,
-            deploy_tier=3
+            deploy_tier=3,
+            groups={"all", "executor"}
         ),
         'control': ServiceConfig(
             gradle_target=':code:services-core:control-service:docker',
             docker_name='control-service',
             instances=None,
-            deploy_tier=0
+            deploy_tier=0,
+            groups={"all", "core"}
         ),
         'query': ServiceConfig(
             gradle_target=':code:services-core:query-service:docker',
             docker_name='query-service',
             instances=2,
-            deploy_tier=2
+            deploy_tier=2,
+            groups={"all", "query"}
         ),
     }
 
     try:
-        tags = get_deployment_tag()
+        parser = argparse.ArgumentParser(
-        if tags == None:
+            prog='deployment.py',
-            exit
+            description='Continuous Deployment helper')
+        parser.add_argument('-v', '--verify', help='Verify the tags are valid, if present', action='store_true')
+        parser.add_argument('-t', '--tag', help='Use the specified tag value instead of the head git tag starting with deploy-')
 
-        print(tags)
+        args = parser.parse_args()
+        tags = args.tag
+        if tags is None:
+            tags = get_deployment_tag()
+        else:
+            tags = tags.split(' ')
 
-        plan = parse_deployment_tags(tags, SERVICE_CONFIG)
-        print("\nDeployment Plan:")
-        print("Services to build:", plan.services_to_build)
-        print("Instances to hold:", plan.instances_to_hold)
 
-        print("\nExecution Plan:")
 
-        build_and_deploy(plan, SERVICE_CONFIG)
+        if tags != None:
+            print("Found deployment tags:", tags)
+
+            plan = parse_deployment_tags(tags, SERVICE_CONFIG)
+
+            print("\nDeployment Plan:")
+            print("Services to build:", plan.services_to_build)
+            print("Instances to deploy:", [container.name for container in plan.instances_to_deploy])
+
+            if not args.verify:
+                print("\nExecution Plan:")
+
+                build_and_deploy(plan, SERVICE_CONFIG)
+        else:
+            print("No tags found")
 
     except ValueError as e:
         print(f"Error: {e}")