(ping) Limit SSL certificate validity dates to a maximum timestamp as permitted by database

This will mean we re-try these as an unencrypted Http connection

code/common/db/resources/db/migration/V25_01_0_003__ping_domains__schema_change.sql

@@ -0,0 +1,5 @@
+-- Add additional summary columns to DOMAIN_SECURITY_EVENTS table
+-- to make it easier to make sense of certificate changes
+
+ALTER TABLE DOMAIN_SECURITY_EVENTS ADD COLUMN CHANGE_SCHEMA ENUM('NONE', 'HTTP_TO_HTTPS', 'HTTPS_TO_HTTP', 'UNKNOWN') NOT NULL DEFAULT 'UNKNOWN';
+OPTIMIZE TABLE DOMAIN_SECURITY_EVENTS;

code/common/service/java/nu/marginalia/service/discovery/ServiceRegistryIf.java

@@ -5,12 +5,10 @@ import nu.marginalia.service.discovery.monitor.ServiceChangeMonitor;
 import nu.marginalia.service.discovery.monitor.ServiceMonitorIf;
 import nu.marginalia.service.discovery.property.ServiceEndpoint;
 import nu.marginalia.service.discovery.property.ServiceKey;
+import org.apache.curator.framework.recipes.locks.InterProcessSemaphoreV2;
 
-import java.util.Collection;
 import java.util.List;
 import java.util.UUID;
-import java.util.function.BiConsumer;
-import java.util.function.Consumer;
 
 import static nu.marginalia.service.discovery.property.ServiceEndpoint.InstanceAddress;
 
@@ -66,6 +64,6 @@ public interface ServiceRegistryIf {
 
     void registerProcess(String processName, int nodeId);
     void deregisterProcess(String processName, int nodeId);
-    void watchProcess(String processName, int nodeId, Consumer<Boolean> callback) throws Exception;
-    void watchProcessAnyNode(String processName, Collection<Integer> nodes, BiConsumer<Boolean, Integer> callback) throws Exception;
+
+    InterProcessSemaphoreV2 getSemaphore(String name, int permits) throws Exception;
 }

code/common/service/java/nu/marginalia/service/discovery/ZkServiceRegistry.java

@@ -6,6 +6,7 @@ import nu.marginalia.service.discovery.monitor.ServiceMonitorIf;
 import nu.marginalia.service.discovery.property.ServiceEndpoint;
 import nu.marginalia.service.discovery.property.ServiceKey;
 import org.apache.curator.framework.CuratorFramework;
+import org.apache.curator.framework.recipes.locks.InterProcessSemaphoreV2;
 import org.apache.curator.utils.ZKPaths;
 import org.apache.zookeeper.CreateMode;
 import org.apache.zookeeper.Watcher;
@@ -13,10 +14,11 @@ import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import java.nio.charset.StandardCharsets;
-import java.util.*;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Random;
+import java.util.UUID;
 import java.util.concurrent.TimeUnit;
-import java.util.function.BiConsumer;
-import java.util.function.Consumer;
 
 import static nu.marginalia.service.discovery.property.ServiceEndpoint.InstanceAddress;
 
@@ -283,60 +285,12 @@ public class ZkServiceRegistry implements ServiceRegistryIf {
     }
 
     @Override
-    public void watchProcess(String processName, int nodeId, Consumer<Boolean> callback) throws Exception {
-        String path = "/process-locks/" + processName + "/" + nodeId;
+    public InterProcessSemaphoreV2 getSemaphore(String name, int permits) {
+        if (stopped)
+            throw new IllegalStateException("Service registry is stopped, cannot get semaphore " + name);
 
-        // first check if the path exists and call the callback accordingly
-
-        if (curatorFramework.checkExists().forPath(path) != null) {
-            callback.accept(true);
-        }
-        else {
-            callback.accept(false);
-        }
-
-        curatorFramework.watchers().add()
-                .usingWatcher((Watcher) change -> {
-                    Watcher.Event.EventType type = change.getType();
-
-                    if (type == Watcher.Event.EventType.NodeCreated) {
-                        callback.accept(true);
-                    }
-                    if (type == Watcher.Event.EventType.NodeDeleted) {
-                        callback.accept(false);
-                    }
-                })
-                .forPath(path);
-
-    }
-
-    @Override
-    public void watchProcessAnyNode(String processName, Collection<Integer> nodes, BiConsumer<Boolean, Integer> callback) throws Exception {
-
-        for (int node : nodes) {
-            String path = "/process-locks/" + processName + "/" + node;
-
-            // first check if the path exists and call the callback accordingly
-            if (curatorFramework.checkExists().forPath(path) != null) {
-                callback.accept(true, node);
-            }
-            else {
-                callback.accept(false, node);
-            }
-
-            curatorFramework.watchers().add()
-                    .usingWatcher((Watcher) change -> {
-                        Watcher.Event.EventType type = change.getType();
-
-                        if (type == Watcher.Event.EventType.NodeCreated) {
-                            callback.accept(true, node);
-                        }
-                        if (type == Watcher.Event.EventType.NodeDeleted) {
-                            callback.accept(false, node);
-                        }
-                    })
-                    .forPath(path);
-        }
+        String path = "/semaphores/" + name;
+        return new InterProcessSemaphoreV2(curatorFramework, path, permits);
     }
 
     /* Exposed for tests */

code/functions/live-capture/build.gradle

@@ -22,6 +22,7 @@ dependencies {
     implementation project(':code:common:db')
     implementation project(':code:libraries:blocking-thread-pool')
     implementation project(':code:libraries:message-queue')
+    implementation project(':code:libraries:domain-lock')
 
     implementation project(':code:execution:api')
     implementation project(':code:processes:crawling-process:ft-content-type')

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

@@ -1,66 +0,0 @@
-package nu.marginalia.rss.svc;
-
-import nu.marginalia.model.EdgeDomain;
-
-import java.util.Map;
-import java.util.concurrent.ConcurrentHashMap;
-import java.util.concurrent.Semaphore;
-
-/** Holds lock objects for each domain, to prevent multiple threads from
- * crawling the same domain at the same time.
- */
-public class DomainLocks {
-    // The locks are stored in a map, with the domain name as the key.  This map will grow
-    // relatively big, but should be manageable since the number of domains is limited to
-    // a few hundred thousand typically.
-    private final Map<String, Semaphore> locks = new ConcurrentHashMap<>();
-
-    /** Returns a lock object corresponding to the given domain.  The object is returned as-is,
-     * and may be held by another thread.  The caller is responsible for locking and  releasing the lock.
-     */
-    public DomainLock lockDomain(EdgeDomain domain) throws InterruptedException {
-        return new DomainLock(domain.toString(),
-                locks.computeIfAbsent(domain.topDomain.toLowerCase(), this::defaultPermits));
-    }
-
-    private Semaphore defaultPermits(String topDomain) {
-        if (topDomain.equals("wordpress.com"))
-            return new Semaphore(16);
-        if (topDomain.equals("blogspot.com"))
-            return new Semaphore(8);
-
-        if (topDomain.equals("neocities.org"))
-            return new Semaphore(4);
-        if (topDomain.equals("github.io"))
-            return new Semaphore(4);
-
-        if (topDomain.equals("substack.com")) {
-            return new Semaphore(1);
-        }
-        if (topDomain.endsWith(".edu")) {
-            return new Semaphore(1);
-        }
-
-        return new Semaphore(2);
-    }
-
-    public static class DomainLock implements AutoCloseable {
-        private final String domainName;
-        private final Semaphore semaphore;
-
-        DomainLock(String domainName, Semaphore semaphore) throws InterruptedException {
-            this.domainName = domainName;
-            this.semaphore = semaphore;
-
-            Thread.currentThread().setName("fetching:" + domainName + " [await domain lock]");
-            semaphore.acquire();
-            Thread.currentThread().setName("fetching:" + domainName);
-        }
-
-        @Override
-        public void close() {
-            semaphore.release();
-            Thread.currentThread().setName("fetching:" + domainName + " [wrapping up]");
-        }
-    }
-}

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

@@ -5,6 +5,8 @@ import com.opencsv.CSVReader;
 import nu.marginalia.WmsaHome;
 import nu.marginalia.contenttype.ContentType;
 import nu.marginalia.contenttype.DocumentBodyToString;
+import nu.marginalia.coordination.DomainCoordinator;
+import nu.marginalia.coordination.DomainLock;
 import nu.marginalia.executor.client.ExecutorClient;
 import nu.marginalia.model.EdgeDomain;
 import nu.marginalia.nodecfg.NodeConfigurationService;
@@ -51,12 +53,13 @@ public class FeedFetcherService {
     private final ServiceHeartbeat serviceHeartbeat;
     private final ExecutorClient executorClient;
 
-    private final DomainLocks domainLocks = new DomainLocks();
+    private final DomainCoordinator domainCoordinator;
 
     private volatile boolean updating;
 
     @Inject
     public FeedFetcherService(FeedDb feedDb,
+                              DomainCoordinator domainCoordinator,
                               FileStorageService fileStorageService,
                               NodeConfigurationService nodeConfigurationService,
                               ServiceHeartbeat serviceHeartbeat,
@@ -67,6 +70,7 @@ public class FeedFetcherService {
         this.nodeConfigurationService = nodeConfigurationService;
         this.serviceHeartbeat = serviceHeartbeat;
         this.executorClient = executorClient;
+        this.domainCoordinator = domainCoordinator;
     }
 
     public enum UpdateMode {
@@ -132,7 +136,7 @@ public class FeedFetcherService {
                         };
 
                         FetchResult feedData;
-                        try (DomainLocks.DomainLock domainLock = domainLocks.lockDomain(new EdgeDomain(feed.domain()))) {
+                        try (DomainLock domainLock = domainCoordinator.lockDomain(new EdgeDomain(feed.domain()))) {
                             feedData = fetchFeedData(feed, client, fetchExecutor, ifModifiedSinceDate, ifNoneMatchTag);
                         } catch (Exception ex) {
                             feedData = new FetchResult.TransientError();

code/libraries/domain-lock/build.gradle

@@ -0,0 +1,32 @@
+plugins {
+    id 'java'
+}
+
+java {
+    toolchain {
+        languageVersion.set(JavaLanguageVersion.of(rootProject.ext.jvmVersion))
+    }
+}
+
+apply from: "$rootProject.projectDir/srcsets.gradle"
+
+dependencies {
+    implementation libs.bundles.slf4j
+    implementation project(':code:common:model')
+    implementation project(':code:common:config')
+    implementation project(':code:common:service')
+
+    implementation libs.bundles.curator
+
+    implementation libs.guava
+    implementation dependencies.create(libs.guice.get()) {
+        exclude group: 'com.google.guava'
+    }
+    testImplementation libs.bundles.slf4j.test
+    testImplementation libs.bundles.junit
+    testImplementation libs.mockito
+}
+
+test {
+    useJUnitPlatform()
+}

code/libraries/domain-lock/java/nu/marginalia/coordination/DefaultDomainPermits.java

@@ -0,0 +1,32 @@
+package nu.marginalia.coordination;
+
+import nu.marginalia.model.EdgeDomain;
+
+public class DefaultDomainPermits {
+
+    public static int defaultPermits(EdgeDomain domain) {
+        return defaultPermits(domain.topDomain.toLowerCase());
+    }
+
+    public static int defaultPermits(String topDomain) {
+
+        if (topDomain.equals("wordpress.com"))
+            return 16;
+        if (topDomain.equals("blogspot.com"))
+            return 8;
+        if (topDomain.equals("tumblr.com"))
+            return 8;
+        if (topDomain.equals("neocities.org"))
+            return 8;
+        if (topDomain.equals("github.io"))
+            return 8;
+        // Substack really dislikes broad-scale crawlers, so we need to be careful
+        // to not get blocked.
+        if (topDomain.equals("substack.com")) {
+            return 1;
+        }
+
+        return 2;
+    }
+
+}

code/libraries/domain-lock/java/nu/marginalia/coordination/DomainCoordinationModule.java

@@ -0,0 +1,17 @@
+package nu.marginalia.coordination;
+
+import com.google.inject.AbstractModule;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+public class DomainCoordinationModule extends AbstractModule {
+
+    private static final Logger logger = LoggerFactory.getLogger(DomainCoordinationModule.class);
+
+    public DomainCoordinationModule() {
+    }
+
+    public void configure() {
+        bind(DomainCoordinator.class).to(ZookeeperDomainCoordinator.class);
+    }
+}

code/libraries/domain-lock/java/nu/marginalia/coordination/DomainCoordinator.java

@@ -0,0 +1,13 @@
+package nu.marginalia.coordination;
+
+import nu.marginalia.model.EdgeDomain;
+
+import java.time.Duration;
+import java.util.Optional;
+
+public interface DomainCoordinator {
+    DomainLock lockDomain(EdgeDomain domain) throws InterruptedException;
+    Optional<DomainLock> tryLockDomain(EdgeDomain domain, Duration timeout) throws InterruptedException;
+    Optional<DomainLock> tryLockDomain(EdgeDomain domain)  throws InterruptedException;
+    boolean isLockableHint(EdgeDomain domain);
+}

code/libraries/domain-lock/java/nu/marginalia/coordination/DomainLock.java

@@ -0,0 +1,5 @@
+package nu.marginalia.coordination;
+
+public interface DomainLock extends AutoCloseable {
+    void close();
+}

code/libraries/domain-lock/java/nu/marginalia/coordination/LocalDomainCoordinator.java

@@ -1,16 +1,17 @@
-package nu.marginalia.crawl.logic;
+package nu.marginalia.coordination;
 
+import com.google.inject.Singleton;
 import nu.marginalia.model.EdgeDomain;
 
+import java.time.Duration;
 import java.util.Map;
 import java.util.Optional;
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.Semaphore;
+import java.util.concurrent.TimeUnit;
 
-/** Holds lock objects for each domain, to prevent multiple threads from
- * crawling the same domain at the same time.
- */
-public class DomainLocks {
+@Singleton
+public class LocalDomainCoordinator implements DomainCoordinator {
     // The locks are stored in a map, with the domain name as the key.  This map will grow
     // relatively big, but should be manageable since the number of domains is limited to
     // a few hundred thousand typically.
@@ -24,13 +25,25 @@ public class DomainLocks {
 
         sem.acquire();
 
-        return new DomainLock(sem);
+        return new LocalDomainLock(sem);
     }
 
     public Optional<DomainLock> tryLockDomain(EdgeDomain domain) {
         var sem = locks.computeIfAbsent(domain.topDomain.toLowerCase(), this::defaultPermits);
         if (sem.tryAcquire(1)) {
-            return Optional.of(new DomainLock(sem));
+            return Optional.of(new LocalDomainLock(sem));
+        }
+        else {
+            // We don't have a lock, so we return an empty optional
+            return Optional.empty();
+        }
+    }
+
+
+    public Optional<DomainLock> tryLockDomain(EdgeDomain domain, Duration timeout) throws InterruptedException {
+        var sem = locks.computeIfAbsent(domain.topDomain.toLowerCase(), this::defaultPermits);
+        if (sem.tryAcquire(1, timeout.toMillis(), TimeUnit.MILLISECONDS)) {
+            return Optional.of(new LocalDomainLock(sem));
         }
         else {
             // We don't have a lock, so we return an empty optional
@@ -39,24 +52,7 @@ public class DomainLocks {
     }
 
     private Semaphore defaultPermits(String topDomain) {
-        if (topDomain.equals("wordpress.com"))
-            return new Semaphore(16);
-        if (topDomain.equals("blogspot.com"))
-            return new Semaphore(8);
-        if (topDomain.equals("tumblr.com"))
-            return new Semaphore(8);
-        if (topDomain.equals("neocities.org"))
-            return new Semaphore(8);
-        if (topDomain.equals("github.io"))
-            return new Semaphore(8);
-
-        // Substack really dislikes broad-scale crawlers, so we need to be careful
-        // to not get blocked.
-        if (topDomain.equals("substack.com")) {
-            return new Semaphore(1);
-        }
-
-        return new Semaphore(2);
+        return new Semaphore(DefaultDomainPermits.defaultPermits(topDomain));
     }
 
     /** Returns true if the domain is lockable, i.e. if it is not already locked by another thread.
@@ -71,15 +67,15 @@ public class DomainLocks {
             return sem.availablePermits() > 0;
     }
 
-    public static class DomainLock implements AutoCloseable {
+    public static class LocalDomainLock implements DomainLock {
         private final Semaphore semaphore;
 
-        DomainLock(Semaphore semaphore) {
+        LocalDomainLock(Semaphore semaphore) {
             this.semaphore = semaphore;
         }
 
         @Override
-        public void close() throws Exception {
+        public void close() {
             semaphore.release();
             Thread.currentThread().setName("[idle]");
         }

code/libraries/domain-lock/java/nu/marginalia/coordination/ZookeeperDomainCoordinator.java

@@ -0,0 +1,116 @@
+package nu.marginalia.coordination;
+
+import com.google.inject.Inject;
+import com.google.inject.Singleton;
+import com.google.inject.name.Named;
+import nu.marginalia.model.EdgeDomain;
+import nu.marginalia.service.discovery.ServiceRegistryIf;
+import org.apache.curator.framework.recipes.locks.InterProcessSemaphoreV2;
+import org.apache.curator.framework.recipes.locks.Lease;
+
+import java.time.Duration;
+import java.util.Map;
+import java.util.Optional;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.TimeUnit;
+
+@Singleton
+public class ZookeeperDomainCoordinator implements DomainCoordinator {
+    // The locks are stored in a map, with the domain name as the key.  This map will grow
+    // relatively big, but should be manageable since the number of domains is limited to
+    // a few hundred thousand typically.
+    private final Map<String, InterProcessSemaphoreV2> locks = new ConcurrentHashMap<>();
+    private final Map<String, Integer> waitCounts = new ConcurrentHashMap<>();
+
+    private final ServiceRegistryIf serviceRegistry;
+    private final int nodeId;
+
+    @Inject
+    public ZookeeperDomainCoordinator(ServiceRegistryIf serviceRegistry, @Named("wmsa-system-node") int nodeId) {
+        // Zookeeper-specific initialization can be done here if needed
+        this.serviceRegistry = serviceRegistry;
+        this.nodeId = nodeId;
+    }
+
+    /** Returns a lock object corresponding to the given domain.  The object is returned as-is,
+     * and may be held by another thread.  The caller is responsible for locking and  releasing the lock.
+     */
+    public DomainLock lockDomain(EdgeDomain domain) throws InterruptedException {
+        final String key = domain.topDomain.toLowerCase();
+        var sem = locks.computeIfAbsent(key, this::createSemapore);
+
+        // Increment or add a wait count for the domain
+        waitCounts.compute(key, (k,value) -> (value == null ? 1 : value + 1));
+        try {
+            return new ZkDomainLock(sem, sem.acquire());
+        }
+        catch (Exception e) {
+            throw new RuntimeException("Failed to acquire lock for domain: " + domain.topDomain, e);
+        }
+        finally {
+            // Decrement or remove the wait count for the domain
+            waitCounts.compute(key, (k,value) -> (value == null || value <= 1) ? null : value - 1);
+        }
+    }
+
+
+    public Optional<DomainLock> tryLockDomain(EdgeDomain domain) throws InterruptedException {
+        return tryLockDomain(domain, Duration.ofSeconds(1)); // Underlying semaphore doesn't have a tryLock method, so we use a short timeout
+    }
+
+
+    public Optional<DomainLock> tryLockDomain(EdgeDomain domain, Duration timeout) throws InterruptedException {
+        final String key = domain.topDomain.toLowerCase();
+        var sem = locks.computeIfAbsent(key, this::createSemapore);
+
+        // Increment or add a wait count for the domain
+        waitCounts.compute(key, (k,value) -> (value == null ? 1 : value + 1));
+        try {
+            var lease = sem.acquire(timeout.toMillis(), TimeUnit.MILLISECONDS); // Acquire with timeout
+            if (lease != null) {
+                return Optional.of(new ZkDomainLock(sem, lease));
+            }
+            else {
+                return Optional.empty(); // If we fail to acquire the lease, we return an empty optional
+            }
+        }
+        catch (Exception e) {
+            return Optional.empty(); // If we fail to acquire the lock, we return an empty optional
+        }
+        finally {
+            waitCounts.compute(key, (k,value) -> (value == null || value <= 1) ? null : value - 1);
+        }
+    }
+
+    private InterProcessSemaphoreV2 createSemapore(String topDomain){
+        try {
+            return serviceRegistry.getSemaphore(topDomain + ":" + nodeId, DefaultDomainPermits.defaultPermits(topDomain));
+        }
+        catch (Exception e) {
+            throw new RuntimeException("Failed to get semaphore for domain: " + topDomain, e);
+        }
+    }
+
+    /** Returns true if the domain is lockable, i.e. if it is not already locked by another thread.
+     * (this is just a hint, and does not guarantee that the domain is actually lockable any time
+     * after this method returns true)
+     */
+    public boolean isLockableHint(EdgeDomain domain) {
+        return !waitCounts.containsKey(domain.topDomain.toLowerCase());
+    }
+
+    public static class ZkDomainLock implements DomainLock {
+        private final InterProcessSemaphoreV2 semaphore;
+        private final Lease lease;
+
+        ZkDomainLock(InterProcessSemaphoreV2 semaphore, Lease lease) {
+            this.semaphore = semaphore;
+            this.lease = lease;
+        }
+
+        @Override
+        public void close() {
+            semaphore.returnLease(lease);
+        }
+    }
+}

code/processes/crawling-process/build.gradle

@@ -32,6 +32,7 @@ dependencies {
     implementation project(':code:libraries:message-queue')
     implementation project(':code:libraries:language-processing')
     implementation project(':code:libraries:easy-lsh')
+    implementation project(':code:libraries:domain-lock')
     implementation project(':code:processes:crawling-process:model')
     implementation project(':code:processes:crawling-process:model')
 

code/processes/crawling-process/java/nu/marginalia/crawl/CrawlerMain.java

@@ -10,9 +10,11 @@ import nu.marginalia.WmsaHome;
 import nu.marginalia.atags.model.DomainLinks;
 import nu.marginalia.atags.source.AnchorTagsSource;
 import nu.marginalia.atags.source.AnchorTagsSourceFactory;
+import nu.marginalia.coordination.DomainCoordinationModule;
+import nu.marginalia.coordination.DomainCoordinator;
+import nu.marginalia.coordination.DomainLock;
 import nu.marginalia.crawl.fetcher.HttpFetcherImpl;
 import nu.marginalia.crawl.fetcher.warc.WarcRecorder;
-import nu.marginalia.crawl.logic.DomainLocks;
 import nu.marginalia.crawl.retreival.CrawlDataReference;
 import nu.marginalia.crawl.retreival.CrawlerRetreiver;
 import nu.marginalia.crawl.retreival.DomainProber;
@@ -68,7 +70,7 @@ public class CrawlerMain extends ProcessMainClass {
     private final ServiceRegistryIf serviceRegistry;
     private final SimpleBlockingThreadPool pool;
 
-    private final DomainLocks domainLocks = new DomainLocks();
+    private final DomainCoordinator domainCoordinator;
 
     private final Map<String, CrawlTask> pendingCrawlTasks = new ConcurrentHashMap<>();
 
@@ -97,6 +99,7 @@ public class CrawlerMain extends ProcessMainClass {
                        WarcArchiverFactory warcArchiverFactory,
                        HikariDataSource dataSource,
                        DomainBlacklist blacklist,
+                       DomainCoordinator domainCoordinator,
                        ServiceRegistryIf serviceRegistry,
                        Gson gson) throws InterruptedException {
 
@@ -114,6 +117,7 @@ public class CrawlerMain extends ProcessMainClass {
         this.blacklist = blacklist;
         this.node = processConfiguration.node();
         this.serviceRegistry = serviceRegistry;
+        this.domainCoordinator = domainCoordinator;
 
         SimpleBlockingThreadPool.ThreadType threadType;
         if (Boolean.getBoolean("crawler.useVirtualThreads")) {
@@ -157,6 +161,7 @@ public class CrawlerMain extends ProcessMainClass {
                     new CrawlerModule(),
                     new ProcessConfigurationModule("crawler"),
                     new ServiceDiscoveryModule(),
+                    new DomainCoordinationModule(),
                     new DatabaseModule(false)
             );
             var crawler = injector.getInstance(CrawlerMain.class);
@@ -451,7 +456,7 @@ public class CrawlerMain extends ProcessMainClass {
         /** Best effort indicator whether we could start this now without getting stuck in
          * DomainLocks purgatory */
         public boolean canRun() {
-            return domainLocks.isLockableHint(new EdgeDomain(domain));
+            return domainCoordinator.isLockableHint(new EdgeDomain(domain));
         }
 
         @Override
@@ -462,7 +467,7 @@ public class CrawlerMain extends ProcessMainClass {
                 return;
             }
 
-            Optional<DomainLocks.DomainLock> lock = domainLocks.tryLockDomain(new EdgeDomain(domain));
+            Optional<DomainLock> lock = domainCoordinator.tryLockDomain(new EdgeDomain(domain));
             // We don't have a lock, so we can't run this task
             // we return to avoid blocking the pool for too long
             if (lock.isEmpty()) {
@@ -470,7 +475,7 @@ public class CrawlerMain extends ProcessMainClass {
                 retryQueue.put(this);
                 return;
             }
-            DomainLocks.DomainLock domainLock = lock.get();
+            DomainLock domainLock = lock.get();
 
             try (domainLock) {
                 Thread.currentThread().setName("crawling:" + domain);

code/processes/crawling-process/java/nu/marginalia/crawl/fetcher/HttpFetcherImpl.java

@@ -36,7 +36,6 @@ import org.apache.hc.core5.http.io.support.ClassicRequestBuilder;
 import org.apache.hc.core5.http.message.MessageSupport;
 import org.apache.hc.core5.http.protocol.HttpContext;
 import org.apache.hc.core5.pool.PoolStats;
-import org.apache.hc.core5.ssl.SSLContextBuilder;
 import org.apache.hc.core5.util.TimeValue;
 import org.apache.hc.core5.util.Timeout;
 import org.jsoup.Jsoup;
@@ -49,15 +48,12 @@ import org.slf4j.MarkerFactory;
 
 import javax.net.ssl.SSLContext;
 import javax.net.ssl.SSLException;
-import javax.net.ssl.TrustManager;
-import javax.net.ssl.X509TrustManager;
 import java.io.IOException;
 import java.net.SocketTimeoutException;
 import java.net.URISyntaxException;
 import java.net.UnknownHostException;
 import java.security.KeyManagementException;
 import java.security.NoSuchAlgorithmException;
-import java.security.cert.X509Certificate;
 import java.time.Duration;
 import java.time.Instant;
 import java.util.*;
@@ -99,42 +95,12 @@ public class HttpFetcherImpl implements HttpFetcher, HttpRequestRetryStrategy {
                 .setValidateAfterInactivity(TimeValue.ofSeconds(5))
                 .build();
 
-        // No-op up front validation of server certificates.
-        //
-        // We will validate certificates later, after the connection is established
-        // as we want to store the certificate chain and validation
-        // outcome to the database.
-
-        var trustMeBro = new X509TrustManager() {
-            private X509Certificate[] lastServerCertChain;
-
-            @Override
-            public void checkClientTrusted(X509Certificate[] chain, String authType) {
-            }
-
-            @Override
-            public void checkServerTrusted(X509Certificate[] chain, String authType) {
-                this.lastServerCertChain = chain.clone();
-            }
-
-            @Override
-            public X509Certificate[] getAcceptedIssuers() {
-                return new X509Certificate[0];
-            }
-
-            public X509Certificate[] getLastServerCertChain() {
-                return lastServerCertChain != null ? lastServerCertChain.clone() : null;
-            }
-        };
-
-        SSLContext sslContext = SSLContextBuilder.create().build();
-        sslContext.init(null, new TrustManager[]{trustMeBro}, null);
 
         connectionManager = PoolingHttpClientConnectionManagerBuilder.create()
                 .setMaxConnPerRoute(2)
                 .setMaxConnTotal(5000)
                 .setDefaultConnectionConfig(connectionConfig)
-                .setTlsSocketStrategy(new DefaultClientTlsStrategy(sslContext))
+                .setTlsSocketStrategy(new DefaultClientTlsStrategy(SSLContext.getDefault()))
                 .build();
 
         connectionManager.setDefaultSocketConfig(SocketConfig.custom()

code/processes/live-crawling-process/build.gradle

@@ -32,6 +32,7 @@ dependencies {
     implementation project(':code:index:api')
     implementation project(':code:processes:process-mq-api')
     implementation project(':code:libraries:message-queue')
+    implementation project(':code:libraries:domain-lock')
     implementation project(':code:libraries:language-processing')
     implementation project(':code:libraries:easy-lsh')
     implementation project(':code:processes:crawling-process')

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

@@ -10,6 +10,8 @@ import nu.marginalia.api.feeds.FeedsClient;
 import nu.marginalia.converting.ConverterModule;
 import nu.marginalia.converting.processor.DomainProcessor;
 import nu.marginalia.converting.writer.ConverterBatchWriter;
+import nu.marginalia.coordination.DomainCoordinationModule;
+import nu.marginalia.coordination.DomainCoordinator;
 import nu.marginalia.db.DbDomainQueries;
 import nu.marginalia.db.DomainBlacklist;
 import nu.marginalia.io.SerializableCrawlDataStream;
@@ -58,6 +60,7 @@ public class LiveCrawlerMain extends ProcessMainClass {
     private final FileStorageService fileStorageService;
     private final KeywordLoaderService keywordLoaderService;
     private final DocumentLoaderService documentLoaderService;
+    private final DomainCoordinator domainCoordinator;
     private final HikariDataSource dataSource;
 
     @Inject
@@ -71,7 +74,7 @@ public class LiveCrawlerMain extends ProcessMainClass {
                            DomainProcessor domainProcessor,
                            FileStorageService fileStorageService,
                            KeywordLoaderService keywordLoaderService,
-                           DocumentLoaderService documentLoaderService, HikariDataSource dataSource)
+                           DocumentLoaderService documentLoaderService, DomainCoordinator domainCoordinator, HikariDataSource dataSource)
             throws Exception
     {
         super(messageQueueFactory, config, gson, LIVE_CRAWLER_INBOX);
@@ -84,6 +87,7 @@ public class LiveCrawlerMain extends ProcessMainClass {
         this.fileStorageService = fileStorageService;
         this.keywordLoaderService = keywordLoaderService;
         this.documentLoaderService = documentLoaderService;
+        this.domainCoordinator = domainCoordinator;
         this.dataSource = dataSource;
 
         domainBlacklist.waitUntilLoaded();
@@ -107,6 +111,7 @@ public class LiveCrawlerMain extends ProcessMainClass {
         try {
             Injector injector = Guice.createInjector(
                     new LiveCrawlerModule(),
+                    new DomainCoordinationModule(), // 2 hours lease timeout is enough for the live crawler
                     new ProcessConfigurationModule("crawler"),
                     new ConverterModule(),
                     new ServiceDiscoveryModule(),
@@ -172,7 +177,7 @@ public class LiveCrawlerMain extends ProcessMainClass {
 
             processHeartbeat.progress(LiveCrawlState.CRAWLING);
 
-            try (SimpleLinkScraper fetcher = new SimpleLinkScraper(dataSet, domainQueries, domainBlacklist);
+            try (SimpleLinkScraper fetcher = new SimpleLinkScraper(dataSet, domainCoordinator, domainQueries, domainBlacklist);
                  var hb = heartbeat.createAdHocTaskHeartbeat("Live Crawling"))
             {
                 for (Map.Entry<String, List<String>> entry : hb.wrap("Fetching", urlsPerDomain.entrySet())) {

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

@@ -5,8 +5,9 @@ import crawlercommons.robots.SimpleRobotRulesParser;
 import nu.marginalia.WmsaHome;
 import nu.marginalia.contenttype.ContentType;
 import nu.marginalia.contenttype.DocumentBodyToString;
+import nu.marginalia.coordination.DomainCoordinator;
+import nu.marginalia.coordination.DomainLock;
 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;
@@ -46,14 +47,16 @@ 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();
+    private final DomainCoordinator domainCoordinator;
 
     private final static int MAX_SIZE = Integer.getInteger("crawler.maxFetchSize", 10 * 1024 * 1024);
 
     public SimpleLinkScraper(LiveCrawlDataSet dataSet,
+                             DomainCoordinator domainCoordinator,
                              DbDomainQueries domainQueries,
                              DomainBlacklist domainBlacklist) {
         this.dataSet = dataSet;
+        this.domainCoordinator = domainCoordinator;
         this.domainQueries = domainQueries;
         this.domainBlacklist = domainBlacklist;
     }
@@ -98,7 +101,7 @@ public class SimpleLinkScraper implements AutoCloseable {
                 .version(HttpClient.Version.HTTP_2)
                 .build();
              // throttle concurrent access per domain; IDE will complain it's not used, but it holds a semaphore -- do not remove:
-             DomainLocks.DomainLock lock = domainLocks.lockDomain(domain)
+             DomainLock lock = domainCoordinator.lockDomain(domain)
         ) {
             SimpleRobotRules rules = fetchRobotsRules(rootUrl, client);
 

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

@@ -1,5 +1,6 @@
 package nu.marginalia.livecrawler;
 
+import nu.marginalia.coordination.LocalDomainCoordinator;
 import nu.marginalia.db.DomainBlacklistImpl;
 import nu.marginalia.io.SerializableCrawlDataStream;
 import nu.marginalia.model.EdgeDomain;
@@ -37,7 +38,7 @@ class SimpleLinkScraperTest {
 
     @Test
     public void testRetrieveNow() throws Exception {
-        var scraper = new SimpleLinkScraper(dataSet, null, Mockito.mock(DomainBlacklistImpl.class));
+        var scraper = new SimpleLinkScraper(dataSet, new LocalDomainCoordinator(),  null, Mockito.mock(DomainBlacklistImpl.class));
         int fetched = scraper.retrieveNow(new EdgeDomain("www.marginalia.nu"), 1, List.of("https://www.marginalia.nu/"));
         Assertions.assertEquals(1, fetched);
 
@@ -57,7 +58,7 @@ class SimpleLinkScraperTest {
     @Test
     public void testRetrieveNow_Redundant() throws Exception {
         dataSet.saveDocument(1, new EdgeUrl("https://www.marginalia.nu/"), "<html>", "", "127.0.0.1");
-        var scraper = new SimpleLinkScraper(dataSet, null, Mockito.mock(DomainBlacklistImpl.class));
+        var scraper = new SimpleLinkScraper(dataSet, new LocalDomainCoordinator(),null, Mockito.mock(DomainBlacklistImpl.class));
 
         // If the requested URL is already in the dataSet, we retrieveNow should shortcircuit and not fetch anything
         int fetched = scraper.retrieveNow(new EdgeDomain("www.marginalia.nu"), 1, List.of("https://www.marginalia.nu/"));

code/processes/ping-process/build.gradle

@@ -27,6 +27,7 @@ dependencies {
     implementation project(':code:common:config')
     implementation project(':code:common:service')
 
+    implementation project(':code:libraries:domain-lock')
     implementation project(':code:libraries:geo-ip')
     implementation project(':code:libraries:message-queue')
 

code/processes/ping-process/java/nu/marginalia/ping/PingDao.java

@@ -185,12 +185,12 @@ public class PingDao {
         return null;
     }
 
-    public List<UpdateSchedule.UpdateJob<Long, HistoricalAvailabilityData>> getDomainUpdateSchedule(int nodeId) {
-        List<UpdateSchedule.UpdateJob<Long, HistoricalAvailabilityData>> updateJobs = new ArrayList<>();
+    public List<UpdateSchedule.UpdateJob<DomainReference, HistoricalAvailabilityData>> getDomainUpdateSchedule(int nodeId) {
+        List<UpdateSchedule.UpdateJob<DomainReference, HistoricalAvailabilityData>> updateJobs = new ArrayList<>();
 
         try (var conn = dataSource.getConnection();
              var ps = conn.prepareStatement("""
-                SELECT ID, NEXT_SCHEDULED_UPDATE
+                SELECT ID, DOMAIN_NAME, NEXT_SCHEDULED_UPDATE
                 FROM EC_DOMAIN
                 LEFT JOIN DOMAIN_AVAILABILITY_INFORMATION
                 ON EC_DOMAIN.ID = DOMAIN_AVAILABILITY_INFORMATION.DOMAIN_ID
@@ -200,11 +200,13 @@ public class PingDao {
             ps.setInt(1, nodeId);
             ResultSet rs = ps.executeQuery();
             while (rs.next()) {
-                long domainId = rs.getLong("ID");
+                int domainId = rs.getInt("ID");
+                String domainName = rs.getString("DOMAIN_NAME");
                 var ts = rs.getTimestamp("NEXT_SCHEDULED_UPDATE");
                 Instant nextUpdate = ts == null ? Instant.now() : ts.toInstant();
 
-                updateJobs.add(new UpdateSchedule.UpdateJob<>(domainId, nextUpdate));
+                var ref = new DomainReference(domainId, nodeId, domainName.toLowerCase());
+                updateJobs.add(new UpdateSchedule.UpdateJob<>(ref, nextUpdate));
             }
         } catch (SQLException e) {
             throw new RuntimeException("Failed to retrieve domain update schedule", e);
@@ -241,7 +243,7 @@ public class PingDao {
                 else {
                     var record = new DomainDnsRecord(rs);
                     updateJobs.add(new UpdateSchedule.UpdateJob<>(
-                            new RootDomainReference.ById(dnsRootDomainId),
+                            new RootDomainReference.ByIdAndName(dnsRootDomainId, rootDomainName),
                             Objects.requireNonNullElseGet(record.tsNextScheduledUpdate(), Instant::now))
                     );
                 }

code/processes/ping-process/java/nu/marginalia/ping/PingJobScheduler.java

@@ -1,6 +1,8 @@
 package nu.marginalia.ping;
 
 import com.google.inject.Inject;
+import nu.marginalia.coordination.DomainCoordinator;
+import nu.marginalia.model.EdgeDomain;
 import nu.marginalia.ping.model.*;
 import nu.marginalia.ping.svc.DnsPingService;
 import nu.marginalia.ping.svc.HttpPingService;
@@ -23,13 +25,14 @@ import java.util.concurrent.TimeUnit;
 public class PingJobScheduler {
     private final HttpPingService httpPingService;
     private final DnsPingService dnsPingService;
+    private final DomainCoordinator domainCoordinator;
     private final PingDao pingDao;
 
     private static final Logger logger = LoggerFactory.getLogger(PingJobScheduler.class);
 
     private static final UpdateSchedule<RootDomainReference, RootDomainReference> dnsUpdateSchedule
             = new UpdateSchedule<>(250_000);
-    private static final UpdateSchedule<Long, HistoricalAvailabilityData> availabilityUpdateSchedule
+    private static final UpdateSchedule<DomainReference, HistoricalAvailabilityData> availabilityUpdateSchedule
             = new UpdateSchedule<>(250_000);
 
     public volatile Instant dnsLastSync = Instant.now();
@@ -43,14 +46,16 @@ public class PingJobScheduler {
     @Inject
     public PingJobScheduler(HttpPingService httpPingService,
                             DnsPingService dnsPingService,
+                            DomainCoordinator domainCoordinator,
                             PingDao pingDao)
     {
         this.httpPingService = httpPingService;
         this.dnsPingService = dnsPingService;
+        this.domainCoordinator = domainCoordinator;
         this.pingDao = pingDao;
     }
 
-    public synchronized void start(boolean startPaused) {
+    public synchronized void start() {
         if (running)
             return;
 
@@ -100,7 +105,7 @@ public class PingJobScheduler {
         logger.info("PingJobScheduler paused");
     }
 
-    public synchronized void resume(int nodeId) {
+    public synchronized void enableForNode(int nodeId) {
         logger.info("Resuming PingJobScheduler for nodeId: {}", nodeId);
         if (this.nodeId != null) {
             logger.warn("Attempted to resume PingJobScheduler with mismatched nodeId: expected {}, got {}", this.nodeId, nodeId);
@@ -134,7 +139,15 @@ public class PingJobScheduler {
                     continue;
                 }
 
-                long nextId = availabilityUpdateSchedule.next();
+                DomainReference ref = availabilityUpdateSchedule.nextIf(domain -> {
+                    EdgeDomain domainObj = new EdgeDomain(domain.domainName());
+                    if (!domainCoordinator.isLockableHint(domainObj)) {
+                        return false; // Skip locked domains
+                    }
+                    return true; // Process this domain
+                });
+
+                long nextId = ref.domainId();
                 var data = pingDao.getHistoricalAvailabilityData(nextId);
                 if (data == null) {
                     logger.warn("No availability data found for ID: {}", nextId);
@@ -159,7 +172,7 @@ public class PingJobScheduler {
                     for (var object : objects) {
                         var ts = object.nextUpdateTime();
                         if (ts != null) {
-                            availabilityUpdateSchedule.add(nextId, ts);
+                            availabilityUpdateSchedule.add(ref, ts);
                             break;
                         }
                     }
@@ -190,7 +203,7 @@ public class PingJobScheduler {
 
                 try {
                     List<WritableModel> objects = switch(ref) {
-                        case RootDomainReference.ById(long id) -> {
+                        case RootDomainReference.ByIdAndName(long id, String name) -> {
                             var oldRecord = Objects.requireNonNull(pingDao.getDomainDnsRecord(id));
                             yield dnsPingService.pingDomain(oldRecord.rootDomainName(), oldRecord);
                         }

code/processes/ping-process/java/nu/marginalia/ping/PingMain.java

@@ -5,15 +5,14 @@ import com.google.inject.Guice;
 import com.google.inject.Inject;
 import com.google.inject.Injector;
 import nu.marginalia.WmsaHome;
+import nu.marginalia.coordination.DomainCoordinationModule;
 import nu.marginalia.geoip.GeoIpDictionary;
 import nu.marginalia.mq.MessageQueueFactory;
 import nu.marginalia.mqapi.ProcessInboxNames;
 import nu.marginalia.mqapi.ping.PingRequest;
-import nu.marginalia.nodecfg.NodeConfigurationService;
 import nu.marginalia.process.ProcessConfiguration;
 import nu.marginalia.process.ProcessConfigurationModule;
 import nu.marginalia.process.ProcessMainClass;
-import nu.marginalia.service.discovery.ServiceRegistryIf;
 import nu.marginalia.service.module.DatabaseModule;
 import nu.marginalia.service.module.ServiceDiscoveryModule;
 import org.slf4j.Logger;
@@ -25,8 +24,6 @@ public class PingMain extends ProcessMainClass {
     private static final Logger log = LoggerFactory.getLogger(PingMain.class);
 
     private final PingJobScheduler pingJobScheduler;
-    private final ServiceRegistryIf serviceRegistry;
-    private final NodeConfigurationService nodeConfigurationService;
     private final int node;
 
     private static final Logger logger = LoggerFactory.getLogger(PingMain.class);
@@ -36,15 +33,11 @@ public class PingMain extends ProcessMainClass {
                     ProcessConfiguration config,
                     Gson gson,
                     PingJobScheduler pingJobScheduler,
-                    ServiceRegistryIf serviceRegistry,
-                    NodeConfigurationService nodeConfigurationService,
                     ProcessConfiguration processConfiguration
                     ) {
         super(messageQueueFactory, config, gson, ProcessInboxNames.PING_INBOX);
 
         this.pingJobScheduler = pingJobScheduler;
-        this.serviceRegistry = serviceRegistry;
-        this.nodeConfigurationService = nodeConfigurationService;
         this.node = processConfiguration.node();
     }
 
@@ -52,7 +45,8 @@ public class PingMain extends ProcessMainClass {
         log.info("Starting PingMain...");
 
         // Start the ping job scheduler
-        pingJobScheduler.start(true);
+        pingJobScheduler.start();
+        pingJobScheduler.enableForNode(node);
 
         log.info("PingMain started successfully.");
     }
@@ -79,6 +73,7 @@ public class PingMain extends ProcessMainClass {
         Injector injector = Guice.createInjector(
                 new PingModule(),
                 new ServiceDiscoveryModule(),
+                new DomainCoordinationModule(),
                 new ProcessConfigurationModule("ping"),
                 new DatabaseModule(false)
         );
@@ -93,7 +88,10 @@ public class PingMain extends ProcessMainClass {
 
         try {
             main.runPrimary();
-            for(;;) main.wait(); // Wait on the object lock to avoid busy-looping
+            for(;;)
+                synchronized (main) { // Wait on the object lock to avoid busy-looping
+                    main.wait();
+                }
         }
         catch (Throwable ex) {
             logger.error("Error running ping process", ex);

code/processes/ping-process/java/nu/marginalia/ping/UpdateSchedule.java

@@ -2,9 +2,8 @@ package nu.marginalia.ping;
 
 import java.time.Duration;
 import java.time.Instant;
-import java.util.Collection;
-import java.util.Comparator;
-import java.util.PriorityQueue;
+import java.util.*;
+import java.util.function.Predicate;
 
 /** In-memory schedule for updates, allowing jobs to be added and processed in order of their scheduled time.
  * This is not a particularly high-performance implementation, but exists to take contention off the database's
@@ -23,6 +22,9 @@ public class UpdateSchedule<T, T2> {
         notifyAll();
     }
 
+    /** Returns the next job in the queue that is due to be processed.
+     * If no jobs are due, it will block until a job is added or a job becomes due.
+     * */
     public synchronized T next() throws InterruptedException {
         while (true) {
             if (updateQueue.isEmpty()) {
@@ -44,6 +46,56 @@ public class UpdateSchedule<T, T2> {
         }
     }
 
+
+    /** Returns the first job in the queue matching the predicate that is not scheduled into the future,
+     * blocking until a job is added or a job becomes due.
+     */
+    public synchronized T nextIf(Predicate<T> predicate) throws InterruptedException {
+        List<UpdateJob<T, T2>> rejectedJobs = new ArrayList<>();
+
+        try {
+            while (true) {
+                if (updateQueue.isEmpty()) {
+                    wait(); // Wait for a new job to be added
+                    continue;
+                }
+
+                UpdateJob<T, T2> job = updateQueue.peek();
+                Instant now = Instant.now();
+
+                if (job.updateTime.isAfter(now)) {
+                    Duration toWait = Duration.between(now, job.updateTime);
+
+                    // Return the rejected jobs to the queue for other threads to process
+                    updateQueue.addAll(rejectedJobs);
+                    if (!rejectedJobs.isEmpty())
+                        notifyAll();
+                    rejectedJobs.clear();
+
+                    wait(Math.max(1, toWait.toMillis()));
+                } else {
+                    var candidate = updateQueue.poll(); // Remove the job from the queue since it's due
+
+                    assert candidate != null : "Update job should not be null at this point, since we just peeked it in a synchronized block";
+
+                    if (!predicate.test(candidate.key())) {
+                        rejectedJobs.add(candidate);
+                    }
+                    else {
+                        return candidate.key();
+                    }
+                }
+            }
+        }
+        finally {
+            // Return the rejected jobs to the queue for other threads to process
+            updateQueue.addAll(rejectedJobs);
+            if (!rejectedJobs.isEmpty())
+                notifyAll();
+        }
+
+    }
+
     public synchronized void clear() {
         updateQueue.clear();
         notifyAll();

code/processes/ping-process/java/nu/marginalia/ping/fetcher/PingHttpFetcher.java

@@ -4,12 +4,14 @@ import com.google.inject.Inject;
 import nu.marginalia.UserAgent;
 import nu.marginalia.WmsaHome;
 import nu.marginalia.ping.fetcher.response.*;
+import org.apache.hc.client5.http.HttpHostConnectException;
 import org.apache.hc.client5.http.classic.HttpClient;
 import org.apache.hc.client5.http.protocol.HttpClientContext;
 import org.apache.hc.core5.http.Header;
 import org.apache.hc.core5.http.io.entity.EntityUtils;
 import org.apache.hc.core5.http.io.support.ClassicRequestBuilder;
 
+import javax.net.ssl.SSLHandshakeException;
 import java.io.IOException;
 import java.net.SocketTimeoutException;
 import java.time.Duration;
@@ -82,9 +84,12 @@ public class PingHttpFetcher {
             });
         } catch (SocketTimeoutException ex) {
             return new TimeoutResponse(ex.getMessage());
-        } catch (IOException e) {
+        } catch (HttpHostConnectException | SSLHandshakeException e) {
             return new ConnectionError(e.getClass().getSimpleName());
+        } catch (IOException e) {
+            return new ProtocolError(e.getClass().getSimpleName());
         }
+
     }
 
 }

code/processes/ping-process/java/nu/marginalia/ping/io/HttpClientProvider.java

@@ -18,13 +18,18 @@ import org.apache.hc.core5.http.HttpResponse;
 import org.apache.hc.core5.http.io.SocketConfig;
 import org.apache.hc.core5.http.message.MessageSupport;
 import org.apache.hc.core5.http.protocol.HttpContext;
+import org.apache.hc.core5.ssl.SSLContextBuilder;
 import org.apache.hc.core5.util.TimeValue;
 import org.apache.hc.core5.util.Timeout;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 import javax.net.ssl.SSLContext;
+import javax.net.ssl.TrustManager;
+import javax.net.ssl.X509TrustManager;
+import java.security.KeyManagementException;
 import java.security.NoSuchAlgorithmException;
+import java.security.cert.X509Certificate;
 import java.util.Iterator;
 import java.util.concurrent.TimeUnit;
 
@@ -37,24 +42,55 @@ public class HttpClientProvider implements Provider<HttpClient> {
     static {
         try {
             client = createClient();
-        } catch (NoSuchAlgorithmException e) {
+        } catch (Exception e) {
             throw new RuntimeException(e);
         }
     }
 
-    private static CloseableHttpClient createClient() throws NoSuchAlgorithmException {
+    private static CloseableHttpClient createClient() throws NoSuchAlgorithmException, KeyManagementException {
         final ConnectionConfig connectionConfig = ConnectionConfig.custom()
                 .setSocketTimeout(15, TimeUnit.SECONDS)
                 .setConnectTimeout(15, TimeUnit.SECONDS)
                 .setValidateAfterInactivity(TimeValue.ofSeconds(5))
                 .build();
 
+        // No-op up front validation of server certificates.
+        //
+        // We will validate certificates later, after the connection is established
+        // as we want to store the certificate chain and validation
+        // outcome to the database.
+
+        var trustMeBro = new X509TrustManager() {
+            private X509Certificate[] lastServerCertChain;
+
+            @Override
+            public void checkClientTrusted(X509Certificate[] chain, String authType) {
+            }
+
+            @Override
+            public void checkServerTrusted(X509Certificate[] chain, String authType) {
+                this.lastServerCertChain = chain.clone();
+            }
+
+            @Override
+            public X509Certificate[] getAcceptedIssuers() {
+                return new X509Certificate[0];
+            }
+
+            public X509Certificate[] getLastServerCertChain() {
+                return lastServerCertChain != null ? lastServerCertChain.clone() : null;
+            }
+        };
+
+        SSLContext sslContext = SSLContextBuilder.create().build();
+        sslContext.init(null, new TrustManager[]{trustMeBro}, null);
+
         connectionManager = PoolingHttpClientConnectionManagerBuilder.create()
                 .setMaxConnPerRoute(2)
                 .setMaxConnTotal(50)
                 .setDefaultConnectionConfig(connectionConfig)
                 .setTlsSocketStrategy(
-                        new DefaultClientTlsStrategy(SSLContext.getDefault(), NoopHostnameVerifier.INSTANCE))
+                        new DefaultClientTlsStrategy(sslContext, NoopHostnameVerifier.INSTANCE))
                 .build();
 
         connectionManager.setDefaultSocketConfig(SocketConfig.custom()

code/processes/ping-process/java/nu/marginalia/ping/io/RetryStrategy.java

@@ -1,5 +1,6 @@
 package nu.marginalia.ping.io;
 
+import org.apache.hc.client5.http.HttpHostConnectException;
 import org.apache.hc.client5.http.HttpRequestRetryStrategy;
 import org.apache.hc.core5.http.HttpRequest;
 import org.apache.hc.core5.http.HttpResponse;
@@ -22,6 +23,7 @@ public class RetryStrategy implements HttpRequestRetryStrategy {
             case SocketTimeoutException ste -> false;
             case SSLException ssle -> false;
             case UnknownHostException uhe -> false;
+            case HttpHostConnectException ex -> executionCount <= 2; // Only retry once for connection errors
             default -> executionCount <= 3;
         };
     }
@@ -50,7 +52,12 @@ public class RetryStrategy implements HttpRequestRetryStrategy {
 
         if (statusCode == 429) {
             // get the Retry-After header
-            String retryAfter = response.getFirstHeader("Retry-After").getValue();
+            var retryAfterHeader = response.getFirstHeader("Retry-After");
+            if (retryAfterHeader == null) {
+                return TimeValue.ofSeconds(3);
+            }
+
+            String retryAfter = retryAfterHeader.getValue();
             if (retryAfter == null) {
                 return TimeValue.ofSeconds(2);
             }

code/processes/ping-process/java/nu/marginalia/ping/model/DomainAvailabilityRecord.java

@@ -1,5 +1,7 @@
 package nu.marginalia.ping.model;
 
+import org.apache.commons.lang3.StringUtils;
+
 import javax.annotation.Nullable;
 import java.sql.Connection;
 import java.sql.ResultSet;
@@ -154,7 +156,7 @@ implements WritableModel
                 ps.setNull(12, java.sql.Types.SMALLINT);
             }
             else {
-                ps.setShort(12, (short) httpResponseTime().toMillis());
+                ps.setInt(12, Math.clamp(httpResponseTime().toMillis(), 0, 0xFFFF)); // "unsigned short" in SQL
             }
 
             if (errorClassification() == null) {
@@ -279,7 +281,7 @@ implements WritableModel
         }
 
         public Builder httpLocation(String httpLocation) {
-            this.httpLocation = httpLocation;
+            this.httpLocation = StringUtils.abbreviate(httpLocation, "...",255);
             return this;
         }
 

code/processes/ping-process/java/nu/marginalia/ping/model/DomainReference.java

@@ -1,3 +1,10 @@
 package nu.marginalia.ping.model;
 
-public record DomainReference(int domainId, int nodeId, String domainName) { }
+import nu.marginalia.model.EdgeDomain;
+
+public record DomainReference(int domainId, int nodeId, String domainName) {
+    public EdgeDomain asEdgeDomain() {
+        return new EdgeDomain(domainName);
+    }
+
+}

code/processes/ping-process/java/nu/marginalia/ping/model/DomainSecurityEvent.java

@@ -18,6 +18,7 @@ public record DomainSecurityEvent(
         boolean certificatePublicKeyChanged,
         boolean certificateSerialNumberChanged,
         boolean certificateIssuerChanged,
+        SchemaChange schemaChange,
         Duration oldCertificateTimeToExpiry,
         boolean securityHeadersChanged,
         boolean ipChanged,
@@ -45,8 +46,9 @@ public record DomainSecurityEvent(
                 security_signature_before,
                 security_signature_after,
                 change_certificate_serial_number,
-                change_certificate_issuer
-                ) VALUES    (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)
+                change_certificate_issuer,
+                change_schema
+                ) VALUES    (?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?,?)
             """))
         {
 
@@ -81,6 +83,7 @@ public record DomainSecurityEvent(
 
             ps.setBoolean(15, certificateSerialNumberChanged());
             ps.setBoolean(16, certificateIssuerChanged());
+            ps.setString(17, schemaChange.name());
 
             ps.executeUpdate();
         }

code/processes/ping-process/java/nu/marginalia/ping/model/DomainSecurityRecord.java

@@ -332,6 +332,8 @@ public record DomainSecurityRecord(
         private String headerXPoweredBy;
         private Instant tsLastUpdate;
 
+        private static Instant MAX_UNIX_TIMESTAMP = Instant.ofEpochSecond(Integer.MAX_VALUE);
+
         public Builder() {
             // Default values for boolean fields
             this.sslCertWildcard = false;
@@ -375,11 +377,17 @@ public record DomainSecurityRecord(
         }
 
         public Builder sslCertNotBefore(Instant sslCertNotBefore) {
+            if (sslCertNotBefore.isAfter(MAX_UNIX_TIMESTAMP)) {
+                sslCertNotBefore = MAX_UNIX_TIMESTAMP;
+            }
             this.sslCertNotBefore = sslCertNotBefore;
             return this;
         }
 
         public Builder sslCertNotAfter(Instant sslCertNotAfter) {
+            if (sslCertNotAfter.isAfter(MAX_UNIX_TIMESTAMP)) {
+                sslCertNotAfter = MAX_UNIX_TIMESTAMP;
+            }
             this.sslCertNotAfter = sslCertNotAfter;
             return this;
         }

code/processes/ping-process/java/nu/marginalia/ping/model/RootDomainReference.java

@@ -1,6 +1,6 @@
 package nu.marginalia.ping.model;
 
 public sealed interface RootDomainReference {
-    record ById(long id) implements RootDomainReference { }
+    record ByIdAndName(long id, String name) implements RootDomainReference { }
     record ByName(String name) implements RootDomainReference { }
 }

code/processes/ping-process/java/nu/marginalia/ping/model/SchemaChange.java

@@ -0,0 +1,12 @@
+package nu.marginalia.ping.model;
+
+public enum SchemaChange {
+    UNKNOWN,
+    NONE,
+    HTTP_TO_HTTPS,
+    HTTPS_TO_HTTP;
+
+    public boolean isSignificant() {
+        return this != NONE && this != UNKNOWN;
+    }
+}

code/processes/ping-process/java/nu/marginalia/ping/model/comparison/SecurityInformationChange.java

@@ -2,6 +2,9 @@ package nu.marginalia.ping.model.comparison;
 
 import nu.marginalia.ping.model.DomainAvailabilityRecord;
 import nu.marginalia.ping.model.DomainSecurityRecord;
+import nu.marginalia.ping.model.HttpSchema;
+import nu.marginalia.ping.model.SchemaChange;
+import org.jetbrains.annotations.NotNull;
 
 import java.time.Duration;
 import java.time.Instant;
@@ -20,7 +23,8 @@ public record SecurityInformationChange(
         Duration oldCertificateTimeToExpiry,
         boolean isSecurityHeadersChanged,
         boolean isIpAddressChanged,
-        boolean isSoftwareHeaderChanged
+        boolean isSoftwareHeaderChanged,
+        SchemaChange schemaChange
 ) {
     public static SecurityInformationChange between(
             DomainSecurityRecord before, DomainAvailabilityRecord availabilityBefore,
@@ -43,9 +47,10 @@ public record SecurityInformationChange(
         );
 
         boolean securityHeadersChanged = before.securityHeadersHash() != after.securityHeadersHash();
-
         boolean softwareChanged = !Objects.equals(before.headerServer(), after.headerServer());
 
+        SchemaChange schemaChange = getSchemaChange(before, after);
+
         // Note we don't include IP address changes in the overall change status,
         // as this is not alone considered a change in security information; we may have
         // multiple IP addresses for a domain, and the IP address may change frequently
@@ -55,7 +60,8 @@ public record SecurityInformationChange(
                 || certificateFingerprintChanged
                 || securityHeadersChanged
                 || certificateProfileChanged
-                || softwareChanged;
+                || softwareChanged
+                || schemaChange.isSignificant();
 
         return new SecurityInformationChange(
                 isChanged,
@@ -69,9 +75,36 @@ public record SecurityInformationChange(
                 oldCertificateTimeToExpiry,
                 securityHeadersChanged,
                 ipChanged,
-                softwareChanged
+                softwareChanged,
+                schemaChange
         );
     }
 
+    private static @NotNull SchemaChange getSchemaChange(DomainSecurityRecord before, DomainSecurityRecord after) {
+        if (before.httpSchema() == null || after.httpSchema() == null) {
+            return SchemaChange.UNKNOWN;
+        }
+
+        boolean beforeIsHttp = before.httpSchema() == HttpSchema.HTTP;
+        boolean afterIsHttp = after.httpSchema() == HttpSchema.HTTP;
+        boolean beforeIsHttps = before.httpSchema() == HttpSchema.HTTPS;
+        boolean afterIsHttps = after.httpSchema() == HttpSchema.HTTPS;
+
+        SchemaChange schemaChange;
+
+        if (beforeIsHttp && afterIsHttp) {
+            schemaChange = SchemaChange.NONE;
+        } else if (beforeIsHttps && afterIsHttps) {
+            schemaChange = SchemaChange.NONE;
+        } else if (beforeIsHttp && afterIsHttps) {
+            schemaChange = SchemaChange.HTTP_TO_HTTPS;
+        } else if (beforeIsHttps && afterIsHttp) {
+            schemaChange = SchemaChange.HTTPS_TO_HTTP;
+        } else {
+            schemaChange = SchemaChange.UNKNOWN;
+        }
+        return schemaChange;
+    }
+
 
 }

code/processes/ping-process/java/nu/marginalia/ping/svc/DomainAvailabilityInformationFactory.java

@@ -96,6 +96,7 @@ public class DomainAvailabilityInformationFactory {
                 .serverIp(address != null ? address.getAddress() : null)
                 .serverIpAsn(getAsn(address))
                 .httpSchema(HttpSchema.HTTP)
+                .httpLocation(rsp.headers().getFirst("Location"))
                 .httpStatus(rsp.httpStatus())
                 .errorClassification(errorClassification)
                 .httpResponseTime(rsp.httpResponseTime())
@@ -164,6 +165,7 @@ public class DomainAvailabilityInformationFactory {
                 .serverIp(address != null ? address.getAddress() : null)
                 .serverIpAsn(getAsn(address))
                 .httpSchema(HttpSchema.HTTPS)
+                .httpLocation(rsp.headers().getFirst("Location"))
                 .httpStatus(rsp.httpStatus())
                 .errorClassification(errorClassification)
                 .httpResponseTime(rsp.httpResponseTime()) // Placeholder, actual timing not implemented

code/processes/ping-process/java/nu/marginalia/ping/svc/DomainSecurityInformationFactory.java

@@ -14,9 +14,7 @@ import java.security.NoSuchAlgorithmException;
 import java.security.cert.CertificateEncodingException;
 import java.security.cert.X509Certificate;
 import java.time.Instant;
-import java.util.HashSet;
-import java.util.Set;
-import java.util.StringJoiner;
+import java.util.*;
 
 public class DomainSecurityInformationFactory {
     private static final Logger logger = LoggerFactory.getLogger(DomainSecurityInformationFactory.class);
@@ -69,8 +67,11 @@ public class DomainSecurityInformationFactory {
         boolean isWildcard = false;
         try {
             if (sslCertificates != null && sslCertificates.length > 0) {
-                for (var sanEntry : sslCertificates[0].getSubjectAlternativeNames()) {
-
+                Collection<List<?>> sans = sslCertificates[0].getSubjectAlternativeNames();
+                if (sans == null) {
+                    sans = Collections.emptyList();
+                }
+                for (var sanEntry : sans) {
 
                     if (sanEntry != null && sanEntry.size() >= 2) {
                         // Check if the SAN entry is a DNS or IP address

code/processes/ping-process/java/nu/marginalia/ping/svc/HttpPingService.java

@@ -2,6 +2,7 @@ package nu.marginalia.ping.svc;
 
 import com.google.inject.Inject;
 import com.google.inject.Singleton;
+import nu.marginalia.coordination.DomainCoordinator;
 import nu.marginalia.ping.fetcher.PingHttpFetcher;
 import nu.marginalia.ping.fetcher.response.*;
 import nu.marginalia.ping.model.*;
@@ -26,6 +27,7 @@ import java.util.List;
 @Singleton
 public class HttpPingService {
 
+    private final DomainCoordinator domainCoordinator;
     private final PingHttpFetcher pingHttpFetcher;
 
     private final DomainAvailabilityInformationFactory domainAvailabilityInformationFactory;
@@ -36,9 +38,11 @@ public class HttpPingService {
 
     @Inject
     public HttpPingService(
+            DomainCoordinator domainCoordinator,
             PingHttpFetcher pingHttpFetcher,
             DomainAvailabilityInformationFactory domainAvailabilityInformationFactory,
             DomainSecurityInformationFactory domainSecurityInformationFactory) throws Exception {
+        this.domainCoordinator = domainCoordinator;
         this.pingHttpFetcher = pingHttpFetcher;
         this.domainAvailabilityInformationFactory = domainAvailabilityInformationFactory;
         this.domainSecurityInformationFactory = domainSecurityInformationFactory;
@@ -59,7 +63,8 @@ public class HttpPingService {
 
     public List<WritableModel> pingDomain(DomainReference domainReference,
                            @Nullable DomainAvailabilityRecord oldPingStatus,
-                           @Nullable DomainSecurityRecord oldSecurityInformation) throws SQLException {
+                           @Nullable DomainSecurityRecord oldSecurityInformation) throws SQLException, InterruptedException {
+
         // First we figure out if the domain maps to an IP address
 
         List<WritableModel> generatedRecords = new ArrayList<>();
@@ -69,8 +74,9 @@ public class HttpPingService {
 
         if (ipAddress.isEmpty()) {
             result = new UnknownHostError();
-        }
-        else {
+        } else {
+            // lock the domain to prevent concurrent pings
+            try (var _ = domainCoordinator.lockDomain(domainReference.asEdgeDomain())) {
                 String url = "https://" + domainReference.domainName() + "/";
                 String alternateUrl = "http://" + domainReference.domainName() + "/";
 
@@ -79,8 +85,7 @@ public class HttpPingService {
                 if (result instanceof HttpsResponse response && shouldTryGET(response.httpStatus())) {
                     sleep(Duration.ofSeconds(2));
                     result = pingHttpFetcher.fetchUrl(url, Method.GET, null, null);
-            }
-            else if (result instanceof ConnectionError) {
+                } else if (result instanceof ConnectionError) {
                     var result2 = pingHttpFetcher.fetchUrl(alternateUrl, Method.HEAD, null, null);
                     if (!(result2 instanceof ConnectionError)) {
                         result = result2;
@@ -90,6 +95,11 @@ public class HttpPingService {
                         result = pingHttpFetcher.fetchUrl(alternateUrl, Method.GET, null, null);
                     }
                 }
+
+                // Add a grace sleep before we yield the semaphore, so that another thread doesn't
+                // immediately hammer the same domain after it's released.
+                sleep(Duration.ofSeconds(1));
+            }
         }
 
 
@@ -135,7 +145,7 @@ public class HttpPingService {
                         domainReference.nodeId(),
                         oldPingStatus,
                         ErrorClassification.HTTP_CLIENT_ERROR,
-                        null);
+                        rsp.errorMessage());
                 newSecurityInformation = null;
             }
             case HttpResponse httpResponse -> {
@@ -286,6 +296,7 @@ public class HttpPingService {
                 change.isCertificatePublicKeyChanged(),
                 change.isCertificateSerialNumberChanged(),
                 change.isCertificateIssuerChanged(),
+                change.schemaChange(),
                 change.oldCertificateTimeToExpiry(),
                 change.isSecurityHeadersChanged(),
                 change.isIpAddressChanged(),

code/processes/ping-process/test/nu/marginalia/ping/AvailabilityJobSchedulerTest.java

@@ -2,6 +2,7 @@ package nu.marginalia.ping;
 
 import com.zaxxer.hikari.HikariConfig;
 import com.zaxxer.hikari.HikariDataSource;
+import nu.marginalia.coordination.LocalDomainCoordinator;
 import nu.marginalia.geoip.GeoIpDictionary;
 import nu.marginalia.ping.fetcher.PingDnsFetcher;
 import nu.marginalia.ping.fetcher.PingHttpFetcher;
@@ -85,11 +86,14 @@ class AvailabilityJobSchedulerTest {
         DomainDnsInformationFactory dnsDomainInformationFactory = new DomainDnsInformationFactory(processConfig, pic);
 
         PingJobScheduler pingJobScheduler = new PingJobScheduler(
-                new HttpPingService(pingHttpFetcher,
+                new HttpPingService(
+                        new LocalDomainCoordinator(),
+                        pingHttpFetcher,
                         new DomainAvailabilityInformationFactory(new GeoIpDictionary(), new BackoffStrategy(pic)),
                         new DomainSecurityInformationFactory()),
                 new DnsPingService(new PingDnsFetcher(List.of("8.8.8.8", "8.8.4.4")),
                         dnsDomainInformationFactory),
+                new LocalDomainCoordinator(),
                 pingDao
         );
 

code/processes/ping-process/test/nu/marginalia/ping/PingDaoTest.java

@@ -320,6 +320,7 @@ class PingDaoTest {
                 true,
                 true,
                 false,
+                SchemaChange.NONE,
                 Duration.ofDays(30),
                 false,
                 false,

code/processes/ping-process/test/nu/marginalia/ping/PingHttpServiceTest.java

@@ -2,6 +2,7 @@ package nu.marginalia.ping;
 
 import com.zaxxer.hikari.HikariConfig;
 import com.zaxxer.hikari.HikariDataSource;
+import nu.marginalia.coordination.LocalDomainCoordinator;
 import nu.marginalia.geoip.GeoIpDictionary;
 import nu.marginalia.ping.fetcher.PingHttpFetcher;
 import nu.marginalia.ping.io.HttpClientProvider;
@@ -63,6 +64,7 @@ class PingHttpServiceTest {
     public void testGetSslInfo() throws Exception {
         var provider = new HttpClientProvider();
         var pingService = new HttpPingService(
+                new LocalDomainCoordinator(),
                 new PingHttpFetcher(provider.get()),
                 new DomainAvailabilityInformationFactory(new GeoIpDictionary(),
                         new BackoffStrategy(PingModule.createPingIntervalsConfiguration())

code/services-core/assistant-service/build.gradle

@@ -37,6 +37,7 @@ dependencies {
     implementation project(':code:functions:domain-info')
     implementation project(':code:functions:domain-info:api')
 
+    implementation project(':code:libraries:domain-lock')
     implementation project(':code:libraries:geo-ip')
     implementation project(':code:libraries:language-processing')
     implementation project(':code:libraries:term-frequency-dict')

code/services-core/assistant-service/java/nu/marginalia/assistant/AssistantMain.java

@@ -5,6 +5,7 @@ import com.google.inject.Inject;
 import com.google.inject.Injector;
 import io.jooby.ExecutionMode;
 import io.jooby.Jooby;
+import nu.marginalia.coordination.DomainCoordinationModule;
 import nu.marginalia.livecapture.LivecaptureModule;
 import nu.marginalia.service.MainClass;
 import nu.marginalia.service.ServiceId;
@@ -29,6 +30,7 @@ public class AssistantMain extends MainClass {
         Injector injector = Guice.createInjector(
                 new AssistantModule(),
                 new LivecaptureModule(),
+                new DomainCoordinationModule(),
                 new ServiceConfigurationModule(ServiceId.Assistant),
                 new ServiceDiscoveryModule(),
                 new DatabaseModule(false)

settings.gradle

@@ -55,6 +55,7 @@ include 'code:libraries:braille-block-punch-cards'
 include 'code:libraries:language-processing'
 include 'code:libraries:term-frequency-dict'
 include 'code:libraries:test-helpers'
+include 'code:libraries:domain-lock'
 
 include 'code:libraries:message-queue'