[GitHub] [sling-org-apache-sling-discovery-oak] rishabhdaim commented on a diff in pull request #13: SLING-10854 : introducing SlingIdCleanupTask to clean up old slingIds…

["rishabhdaim (via GitHub)" <gi...@apache.org>]

rishabhdaim commented on code in PR #13:
URL: https://github.com/apache/sling-org-apache-sling-discovery-oak/pull/13#discussion_r1133435308


##########
src/main/java/org/apache/sling/discovery/oak/SlingIdCleanupTask.java:
##########
@@ -0,0 +1,601 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied.  See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+package org.apache.sling.discovery.oak;
+
+import static org.osgi.util.converter.Converters.standardConverter;
+
+import java.util.Calendar;
+import java.util.Date;
+import java.util.HashSet;
+import java.util.Set;
+import java.util.UUID;
+import java.util.concurrent.atomic.AtomicInteger;
+
+import org.apache.sling.api.resource.LoginException;
+import org.apache.sling.api.resource.ModifiableValueMap;
+import org.apache.sling.api.resource.PersistenceException;
+import org.apache.sling.api.resource.Resource;
+import org.apache.sling.api.resource.ResourceResolver;
+import org.apache.sling.api.resource.ResourceResolverFactory;
+import org.apache.sling.api.resource.ValueMap;
+import org.apache.sling.commons.scheduler.ScheduleOptions;
+import org.apache.sling.commons.scheduler.Scheduler;
+import org.apache.sling.discovery.InstanceDescription;
+import org.apache.sling.discovery.TopologyEvent;
+import org.apache.sling.discovery.TopologyEvent.Type;
+import org.apache.sling.discovery.TopologyEventListener;
+import org.apache.sling.discovery.TopologyView;
+import org.osgi.framework.BundleContext;
+import org.osgi.service.component.annotations.Activate;
+import org.osgi.service.component.annotations.Component;
+import org.osgi.service.component.annotations.Deactivate;
+import org.osgi.service.component.annotations.Modified;
+import org.osgi.service.component.annotations.Reference;
+import org.osgi.service.metatype.annotations.AttributeDefinition;
+import org.osgi.service.metatype.annotations.Designate;
+import org.osgi.service.metatype.annotations.ObjectClassDefinition;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+
+/**
+ * A background task that cleans up garbage slingIds after topology changes.
+ * <p>
+ * A slingId is considered garbage when:
+ * <ul>
+ * <li>it is not in the current topology</li>
+ * <li>was not ever seen in previous topologies by the now leader instance</li>
+ * <li>it is not in the current idmap (where clusterNodeIds are reused hence
+ * that list stays small and does not need cleanup)</li>
+ * <li>its leaderElectionId was created more than 7 days ago (the
+ * leaderElectionId is created at activate time of the discovery.oak bundle -
+ * hence this more or less corresponds to the startup time of that
+ * instance)</li>
+ * </ul>
+ * The garbage accumulates at the following places, where it will thus be
+ * cleaned up from:
+ * <ul>
+ * <li>as child node under /var/discovery/oak/clusterInstances : this is the
+ * most performance critical garbage</li>
+ * <li>as a property key in /var/discovery/oak/syncTokens</li>
+ * </ul>
+ * The task by default is executed:
+ * <ul>
+ * <li>only on the leader</li>
+ * <li>10min after a TOPOLOGY_INIT or TOPOLOGY_CHANGED event</li>
+ * <li>with a maximum number of delete operations to avoid repository overload -
+ * that maximum is called batchSize and is 50 by default</li>
+ * <li>in subsequent intervals of 10min after the initial run, if that had to
+ * stop at the batchSize of 50 deletions</li>
+ * </ul>
+ * All parameters mentioned above can be configured.
+ * <p>
+ * Additionally, the cleanup is skipped for 13 hours after a successful cleanup.
+ * This is to avoid unnecessary load on the repository. The number of 13
+ * incorporates some heuristics such as : about 2 cleanup rounds per day maximum
+ * makes sense, if a leader is very long living, then the 1 additional hour
+ * makes it spread somewhat throughout the day. This is to further minimize any
+ * load side-effects.
+ */
+@Component
+@Designate(ocd = SlingIdCleanupTask.Conf.class)
+public class SlingIdCleanupTask implements TopologyEventListener, Runnable {
+
+    final static String SLINGID_CLEANUP_ENABLED_SYSTEM_PROPERTY_NAME = "org.apache.sling.discovery.oak.slingidcleanup.enabled";
+
+    /** default minimal cleanup delay at 13h, to intraday load balance */
+    final static long MIN_CLEANUP_DELAY_MILLIS = 46800000;
+
+    /**
+     * default age is 1 week : an instance that is not in the current topology,
+     * started 1 week ago is very unlikely to still be active
+     */
+    private static final long DEFAULT_MIN_CREATION_AGE_MILLIS = 604800000; // 1 week
+
+    /**
+     * initial delay is 10min : after a TOPOLOGY_INIT or TOPOLOGY_CHANGED on the
+     * leader, there should be a 10min delay before starting a round of cleanup.
+     * This is to not add unnecessary load after a startup/change.
+     */
+    private static final int DEFAULT_CLEANUP_INITIAL_DELAY = 600000; // 10min
+
+    /**
+     * default cleanup interval is 10min - this is together with the batchSize to
+     * lower repository load
+     */
+    private static final int DEFAULT_CLEANUP_INTERVAL = 600000; // 10min
+
+    /**
+     * default batch size is 50 deletions : normally there should not be much
+     * garbage around anyway, so normally it's just a few, 1-5 perhaps. If there's
+     * more than 50, that is probably a one-time cleanup after this feature is first
+     * rolled out. That one-time cleanup can actually take a considerable amount of
+     * time. So, to not overload the write load on the repository, the deletion is
+     * batched into 50 at any time - with 10min delays in between. That results in
+     * an average of 1 cleanup every 12 seconds, or 5 per minute, or 8640 per day,
+     * for a legacy cleanup.
+     */
+    private static final int DEFAULT_CLEANUP_BATCH_SIZE = 50;
+
+    /**
+     * The sling.commons.scheduler name, so that it can be cancelled upon topology
+     * changes.
+     */
+    private static final String SCHEDULE_NAME = "org.apache.sling.discovery.oak.SlingIdCleanupTask";
+
+    protected final Logger logger = LoggerFactory.getLogger(this.getClass());
+
+    @Reference
+    protected Scheduler scheduler;
+
+    @Reference
+    protected ResourceResolverFactory resourceResolverFactory;
+
+    @Reference
+    private Config config;
+
+    /**
+     * volatile flag to fast stop any ongoing deletion upon a change in the topology
+     */
+    private volatile boolean hasTopology = false;
+
+    /**
+     * volatile field to keep track of the current topology, shared between topology
+     * listener and deletion
+     */
+    @SuppressWarnings("all")
+    private volatile TopologyView currentView;
+
+    private int initialDelayMillis = DEFAULT_CLEANUP_INITIAL_DELAY;
+
+    private int intervalMillis = DEFAULT_CLEANUP_INTERVAL;
+
+    private int batchSize = DEFAULT_CLEANUP_BATCH_SIZE;
+
+    private long minCreationAgeMillis = DEFAULT_MIN_CREATION_AGE_MILLIS;
+
+    /** test counter that increments upon every scheduler invocation */
+    private AtomicInteger runCount = new AtomicInteger(0);
+
+    /** test counter that increments upon every batch deletion */
+    private AtomicInteger completionCount = new AtomicInteger(0);
+
+    /** test counter that keeps track of actually deleted slingIds */
+    private AtomicInteger deleteCount = new AtomicInteger(0);
+
+    /**
+     * flag to distinguish first from subsequent runs, as they might have different
+     * scheduler delays
+     */
+    private volatile boolean firstRun = true;
+
+    private long lastSuccessfulRun = -1;
+
+    /**
+     * Minimal delay after a successful cleanup round, in millis
+     */
+    private long minCleanupDelayMillis = MIN_CLEANUP_DELAY_MILLIS;
+
+    /**
+     * contains all slingIds ever seen by this instance - should not be a long list
+     * so not a memory issue
+     */
+    private Set<String> seenInstances = new HashSet<>();
+
+    @ObjectClassDefinition(name = "Apache Sling Discovery Oak SlingId Cleanup Task", description = "This task is in charge of cleaning up old SlingIds from the repository.")
+    public @interface Conf {
+
+        @AttributeDefinition(name = "Cleanup initial delay milliseconds", description = "Number of milliseconds to initially wait for the first cleanup")
+        int org_apache_sling_discovery_oak_slingid_cleanup_initial_delay() default DEFAULT_CLEANUP_INITIAL_DELAY;

Review Comment:
   aren't these method names too big?
   
   the part `org_apache_sling_discovery_oak_` can be removed.
   
   For reference, https://github.com/apache/sling-org-apache-sling-discovery-oak/blob/master/src/main/java/org/apache/sling/discovery/oak/DiscoveryServiceCentralConfig.java#L29 doesn't use them.



-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: dev-unsubscribe@sling.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org