avijayanhwx commented on a change in pull request #1456: URL: https://github.com/apache/hadoop-ozone/pull/1456#discussion_r503538733
########## File path: hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/upgrade/OMUpgradeFinalizer.java ########## @@ -0,0 +1,328 @@ +/** + * 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.hadoop.ozone.om.upgrade; + +import org.apache.hadoop.ozone.om.OzoneManager; +import org.apache.hadoop.ozone.om.exceptions.OMException; +import org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes; +import org.apache.hadoop.ozone.upgrade.UpgradeFinalizer; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.List; +import java.util.Queue; +import java.util.concurrent.Callable; +import java.util.concurrent.ConcurrentLinkedQueue; + +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.*; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.INVALID_REQUEST; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.PERSIST_UPGRADE_TO_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.REMOVE_UPGRADE_TO_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.UPDATE_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.upgrade.UpgradeFinalizer.Status.FINALIZATION_DONE; +import static org.apache.hadoop.ozone.upgrade.UpgradeFinalizer.Status.FINALIZATION_IN_PROGRESS; + +/** + * UpgradeFinalizer implementation for the Ozone Manager service. + */ +public class OMUpgradeFinalizer implements UpgradeFinalizer<OzoneManager> { + + private static final OmUpgradeAction NOOP = a -> {}; + + private OMLayoutVersionManagerImpl versionManager; + private String clientID; + + private Queue<String> msgs = new ConcurrentLinkedQueue<>(); + private boolean isDone = false; + + public OMUpgradeFinalizer(OMLayoutVersionManagerImpl versionManager) { + this.versionManager = versionManager; + } + + @Override + public StatusAndMessages finalize(String upgradeClientID, OzoneManager om) + throws IOException { + if (!versionManager.needsFinalization()) { + return FINALIZED_MSG; + } + clientID = upgradeClientID; + +// This requires some more investigation on how to do it properly while +// requests are on the fly, and post finalize features one by one. +// Until that is done, monitoring is not really doing anything meaningful +// but this is a tradoff we can take for the first iteration either if needed, +// as the finalization of the first few features should not take that long. +// Follow up JIRA is in HDDS-4286 +// String threadName = "OzoneManager-Upgrade-Finalizer"; +// ExecutorService executor = +// Executors.newSingleThreadExecutor(r -> new Thread(threadName)); +// executor.submit(new Worker(om)); + new Worker(om).call(); + return STARTING_MSG; + } + + @Override + public synchronized StatusAndMessages reportStatus( + String upgradeClientID, boolean takeover + ) throws IOException { + if (takeover) { + clientID = upgradeClientID; + } + assertClientId(upgradeClientID); + List<String> returningMsgs = new ArrayList<>(msgs.size()+10); + Status status = isDone ? FINALIZATION_DONE : FINALIZATION_IN_PROGRESS; + while (msgs.size() > 0) { + returningMsgs.add(msgs.poll()); + } + return new StatusAndMessages(status, returningMsgs); + } + + private void assertClientId(String id) throws OMException { + if (!this.clientID.equals(id)) { + throw new OMException("Unknown client tries to get finalization status.\n" + + "The requestor is not the initiating client of the finalization," + + " if you want to take over, and get unsent status messages, check" + + " -takeover option.", INVALID_REQUEST); + } + } + + /** + * This class implements the finalization logic applied to every + * LayoutFeature that needs to be finalized. + * + * For the first approach this happens synchronously within the state machine + * during the FinalizeUpgrade request, but ideally this has to be moved to + * individual calls that are going into the StateMaching one by one. + * The prerequisits for this to happen in the background are the following: + * - more fine grained control for LayoutFeatures to prepare the + * finalization outside the state machine, do the switch from old to new + * logic inside the statemachine and apply the finalization, and then do + * any cleanup necessary outside the state machine + * - a way to post a request to the state machine that is not part of the + * client API, so basically not an OMRequest, but preferably an internal + * request, which is posted from the leader OM to the follower OMs only. + * - ensure that there is a possibility to implement a rollback logic if + * something goes wrong inside the state machine, to avoid OM stuck in an + * intermediate state due to an error. + */ + private class Worker implements Callable<Void> { + private OzoneManager ozoneManager; + + /** + * Initiates the Worker, for the specified OM instance. + * @param om the OzoneManager instance on which to finalize the new + * LayoutFeatures. + */ + Worker(OzoneManager om) { + ozoneManager = om; + } + + @Override + public Void call() throws OMException { + try { + emitStartingMsg(); + + for (OMLayoutFeature f : versionManager.unfinalizedFeatures()) { + finalizeFeature(f); + updateLayoutVersionInVersionFile(f); + versionManager.finalized(f); + } + + emitFinishedMsg(); + } finally { + isDone = true; + } + return null; + } + + private void finalizeFeature(OMLayoutFeature feature) + throws OMException { + OmUpgradeAction action = feature.onFinalizeAction().orElse(NOOP); + + if (action == NOOP) { + emitNOOPMsg(feature.name()); + return; + } + + putFinalizationMarkIntoVersionFile(feature); + + emitStartingFinalizationActionMsg(feature.name()); + try { + action.executeAction(ozoneManager); + } catch (Exception e) { + logFinalizationFailureAndThrow(e, feature.name()); + } + emitFinishFinalizationActionMsg(feature.name()); + + removeFinalizationMarkFromVersionFile(feature); + } + + private void updateLayoutVersionInVersionFile(OMLayoutFeature feature) + throws OMException { + int prevLayoutVersion = currentStoredLayoutVersion(); + + updateStorageLayoutVersion(feature.layoutVersion()); + try { + persistStorage(); + } catch (IOException e) { + updateStorageLayoutVersion(prevLayoutVersion); + logLayoutVersionUpdateFailureAndThrow(e); + } + } + + private void putFinalizationMarkIntoVersionFile(OMLayoutFeature feature) + throws OMException { + try { + emitUpgradeToLayoutVersionPersistingMsg(feature.name()); + + setUpgradeToLayoutVersionInStorage(feature.layoutVersion()); + persistStorage(); + + emitUpgradeToLayoutVersionPersistedMsg(); + } catch (IOException e) { + logUpgradeToLayoutVersionPersistingFailureAndThrow(feature.name(), e); + } + } + + private void removeFinalizationMarkFromVersionFile(OMLayoutFeature feature) + throws OMException { + try { + emitRemovingUpgradeToLayoutVersionMsg(feature.name()); + + unsetUpgradeToLayoutVersionInStorage(); + persistStorage(); + + emitRemovedUpgradeToLayoutVersionMsg(); + } catch (IOException e) { + logUpgradeToLayoutVersionRemovalFailureAndThrow(feature.name(), e); + } + } + + + + + + private void setUpgradeToLayoutVersionInStorage(int version) { + ozoneManager.getOmStorage().setUpgradeToLayoutVersion(version); + } + + private void unsetUpgradeToLayoutVersionInStorage() { + ozoneManager.getOmStorage().unsetUpgradeToLayoutVersion(); + } + + private int currentStoredLayoutVersion() { + return ozoneManager.getOmStorage().getLayoutVersion(); + } + + private void updateStorageLayoutVersion(int version) { + ozoneManager.getOmStorage().setLayoutVersion(version); + } + + private void persistStorage() throws IOException { Review comment: Can this be moved to the individual layout version update methods above, so that the caller does not need to invoke the persist explicitly every time? As a caller, if I do **storage**.updateStorageLayoutVersion(), I can expect it to be persisted, right? I don't have a strong opinion about this, just a suggestion. ########## File path: hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/request/upgrade/OMFinalizeUpgradeRequest.java ########## @@ -63,11 +64,20 @@ public OMClientResponse validateAndUpdateCache( String upgradeClientID = request.getUpgradeClientId(); - UpgradeFinalizationStatus status = + StatusAndMessages omStatus = Review comment: Line **54** has a debug log line left. ########## File path: hadoop-ozone/ozone-manager/src/main/java/org/apache/hadoop/ozone/om/upgrade/OMUpgradeFinalizer.java ########## @@ -0,0 +1,328 @@ +/** + * 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.hadoop.ozone.om.upgrade; + +import org.apache.hadoop.ozone.om.OzoneManager; +import org.apache.hadoop.ozone.om.exceptions.OMException; +import org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes; +import org.apache.hadoop.ozone.upgrade.UpgradeFinalizer; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.List; +import java.util.Queue; +import java.util.concurrent.Callable; +import java.util.concurrent.ConcurrentLinkedQueue; + +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.*; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.INVALID_REQUEST; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.PERSIST_UPGRADE_TO_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.REMOVE_UPGRADE_TO_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.om.exceptions.OMException.ResultCodes.UPDATE_LAYOUT_VERSION_FAILED; +import static org.apache.hadoop.ozone.upgrade.UpgradeFinalizer.Status.FINALIZATION_DONE; +import static org.apache.hadoop.ozone.upgrade.UpgradeFinalizer.Status.FINALIZATION_IN_PROGRESS; + +/** + * UpgradeFinalizer implementation for the Ozone Manager service. + */ +public class OMUpgradeFinalizer implements UpgradeFinalizer<OzoneManager> { + + private static final OmUpgradeAction NOOP = a -> {}; + + private OMLayoutVersionManagerImpl versionManager; + private String clientID; + + private Queue<String> msgs = new ConcurrentLinkedQueue<>(); + private boolean isDone = false; + + public OMUpgradeFinalizer(OMLayoutVersionManagerImpl versionManager) { + this.versionManager = versionManager; + } + + @Override + public StatusAndMessages finalize(String upgradeClientID, OzoneManager om) + throws IOException { + if (!versionManager.needsFinalization()) { + return FINALIZED_MSG; + } + clientID = upgradeClientID; + +// This requires some more investigation on how to do it properly while +// requests are on the fly, and post finalize features one by one. +// Until that is done, monitoring is not really doing anything meaningful +// but this is a tradoff we can take for the first iteration either if needed, +// as the finalization of the first few features should not take that long. +// Follow up JIRA is in HDDS-4286 +// String threadName = "OzoneManager-Upgrade-Finalizer"; +// ExecutorService executor = +// Executors.newSingleThreadExecutor(r -> new Thread(threadName)); +// executor.submit(new Worker(om)); + new Worker(om).call(); + return STARTING_MSG; Review comment: Just an observation. By the time we return 'STARTING_MSG' at 77, we are already done with Finalization right? Is this return message left as it is for a future refactoring through HDDS-3286 where async finalization will be done? ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: ozone-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: ozone-issues-h...@hadoop.apache.org
