View Javadoc

1   /**
2    * Licensed to the Apache Software Foundation (ASF) under one
3    * or more contributor license agreements.  See the NOTICE file
4    * distributed with this work for additional information
5    * regarding copyright ownership.  The ASF licenses this file
6    * to you under the Apache License, Version 2.0 (the
7    * "License"); you may not use this file except in compliance
8    * with the License.  You may obtain a copy of the License at
9    *
10   *     http://www.apache.org/licenses/LICENSE-2.0
11   *
12   * Unless required by applicable law or agreed to in writing, software
13   * distributed under the License is distributed on an "AS IS" BASIS,
14   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
15   * See the License for the specific language governing permissions and
16   * limitations under the License.
17   */
18  package org.apache.hadoop.hbase.master.snapshot;
19  
20  import java.io.IOException;
21  import java.util.List;
22  import java.util.Map;
23  import java.util.Set;
24  
25  import org.apache.commons.logging.Log;
26  import org.apache.commons.logging.LogFactory;
27  import org.apache.hadoop.hbase.classification.InterfaceAudience;
28  import org.apache.hadoop.hbase.classification.InterfaceStability;
29  import org.apache.hadoop.fs.FileSystem;
30  import org.apache.hadoop.fs.Path;
31  import org.apache.hadoop.hbase.TableName;
32  import org.apache.hadoop.hbase.HRegionInfo;
33  import org.apache.hadoop.hbase.HTableDescriptor;
34  import org.apache.hadoop.hbase.client.RegionReplicaUtil;
35  import org.apache.hadoop.hbase.MetaTableAccessor;
36  import org.apache.hadoop.hbase.master.MasterServices;
37  import org.apache.hadoop.hbase.mob.MobUtils;
38  import org.apache.hadoop.hbase.protobuf.generated.HBaseProtos.SnapshotDescription;
39  import org.apache.hadoop.hbase.protobuf.generated.SnapshotProtos.SnapshotRegionManifest;
40  import org.apache.hadoop.hbase.snapshot.ClientSnapshotDescriptionUtils;
41  import org.apache.hadoop.hbase.snapshot.CorruptedSnapshotException;
42  import org.apache.hadoop.hbase.snapshot.SnapshotDescriptionUtils;
43  import org.apache.hadoop.hbase.snapshot.SnapshotManifest;
44  import org.apache.hadoop.hbase.snapshot.SnapshotReferenceUtil;
45  import org.apache.hadoop.hbase.zookeeper.MetaTableLocator;
46  
47  /**
48   * General snapshot verification on the master.
49   * <p>
50   * This is a light-weight verification mechanism for all the files in a snapshot. It doesn't
51   * attempt to verify that the files are exact copies (that would be paramount to taking the
52   * snapshot again!), but instead just attempts to ensure that the files match the expected
53   * files and are the same length.
54   * <p>
55   * Taking an online snapshots can race against other operations and this is an last line of
56   * defense.  For example, if meta changes between when snapshots are taken not all regions of a
57   * table may be present.  This can be caused by a region split (daughters present on this scan,
58   * but snapshot took parent), or move (snapshots only checks lists of region servers, a move could
59   * have caused a region to be skipped or done twice).
60   * <p>
61   * Current snapshot files checked:
62   * <ol>
63   * <li>SnapshotDescription is readable</li>
64   * <li>Table info is readable</li>
65   * <li>Regions</li>
66   * </ol>
67   * <ul>
68   * <li>Matching regions in the snapshot as currently in the table</li>
69   * <li>{@link HRegionInfo} matches the current and stored regions</li>
70   * <li>All referenced hfiles have valid names</li>
71   * <li>All the hfiles are present (either in .archive directory in the region)</li>
72   * <li>All recovered.edits files are present (by name) and have the correct file size</li>
73   * </ul>
74   */
75  @InterfaceAudience.Private
76  @InterfaceStability.Unstable
77  public final class MasterSnapshotVerifier {
78    private static final Log LOG = LogFactory.getLog(MasterSnapshotVerifier.class);
79  
80    private SnapshotDescription snapshot;
81    private FileSystem fs;
82    private Path rootDir;
83    private TableName tableName;
84    private MasterServices services;
85  
86    /**
87     * @param services services for the master
88     * @param snapshot snapshot to check
89     * @param rootDir root directory of the hbase installation.
90     */
91    public MasterSnapshotVerifier(MasterServices services, SnapshotDescription snapshot, Path rootDir) {
92      this.fs = services.getMasterFileSystem().getFileSystem();
93      this.services = services;
94      this.snapshot = snapshot;
95      this.rootDir = rootDir;
96      this.tableName = TableName.valueOf(snapshot.getTable());
97    }
98  
99    /**
100    * Verify that the snapshot in the directory is a valid snapshot
101    * @param snapshotDir snapshot directory to check
102    * @param snapshotServers {@link org.apache.hadoop.hbase.ServerName} of the servers 
103    *        that are involved in the snapshot
104    * @throws CorruptedSnapshotException if the snapshot is invalid
105    * @throws IOException if there is an unexpected connection issue to the filesystem
106    */
107   public void verifySnapshot(Path snapshotDir, Set<String> snapshotServers)
108       throws CorruptedSnapshotException, IOException {
109     SnapshotManifest manifest = SnapshotManifest.open(services.getConfiguration(), fs,
110                                                       snapshotDir, snapshot);
111     // verify snapshot info matches
112     verifySnapshotDescription(snapshotDir);
113 
114     // check that tableinfo is a valid table description
115     verifyTableInfo(manifest);
116 
117     // check that each region is valid
118     verifyRegions(manifest);
119   }
120 
121   /**
122    * Check that the snapshot description written in the filesystem matches the current snapshot
123    * @param snapshotDir snapshot directory to check
124    */
125   private void verifySnapshotDescription(Path snapshotDir) throws CorruptedSnapshotException {
126     SnapshotDescription found = SnapshotDescriptionUtils.readSnapshotInfo(fs, snapshotDir);
127     if (!this.snapshot.equals(found)) {
128       throw new CorruptedSnapshotException("Snapshot read (" + found
129           + ") doesn't equal snapshot we ran (" + snapshot + ").", snapshot);
130     }
131   }
132 
133   /**
134    * Check that the table descriptor for the snapshot is a valid table descriptor
135    * @param manifest snapshot manifest to inspect
136    */
137   private void verifyTableInfo(final SnapshotManifest manifest) throws IOException {
138     HTableDescriptor htd = manifest.getTableDescriptor();
139     if (htd == null) {
140       throw new CorruptedSnapshotException("Missing Table Descriptor", snapshot);
141     }
142 
143     if (!htd.getNameAsString().equals(snapshot.getTable())) {
144       throw new CorruptedSnapshotException("Invalid Table Descriptor. Expected "
145         + snapshot.getTable() + " name, got " + htd.getNameAsString(), snapshot);
146     }
147   }
148 
149   /**
150    * Check that all the regions in the snapshot are valid, and accounted for.
151    * @param manifest snapshot manifest to inspect
152    * @throws IOException if we can't reach hbase:meta or read the files from the FS
153    */
154   private void verifyRegions(final SnapshotManifest manifest) throws IOException {
155     List<HRegionInfo> regions;
156     if (TableName.META_TABLE_NAME.equals(tableName)) {
157       regions = new MetaTableLocator().getMetaRegions(services.getZooKeeper());
158     } else {
159       regions = MetaTableAccessor.getTableRegions(services.getConnection(), tableName);
160     }
161     // Remove the non-default regions
162     RegionReplicaUtil.removeNonDefaultRegions(regions);
163 
164     Map<String, SnapshotRegionManifest> regionManifests = manifest.getRegionManifestsMap();
165     if (regionManifests == null) {
166       String msg = "Snapshot " + ClientSnapshotDescriptionUtils.toString(snapshot) + " looks empty";
167       LOG.error(msg);
168       throw new CorruptedSnapshotException(msg);
169     }
170 
171     String errorMsg = "";
172     boolean hasMobStore = false;
173     // the mob region is a dummy region, it's not a real region in HBase.
174     // the mob region has a special name, it could be found by the region name.
175     if (regionManifests.get(MobUtils.getMobRegionInfo(tableName).getEncodedName()) != null) {
176       hasMobStore = true;
177     }
178     int realRegionCount = hasMobStore ? regionManifests.size() - 1 : regionManifests.size();
179     if (realRegionCount != regions.size()) {
180       errorMsg = "Regions moved during the snapshot '" +
181                    ClientSnapshotDescriptionUtils.toString(snapshot) + "'. expected=" +
182                    regions.size() + " snapshotted=" + realRegionCount + ".";
183       LOG.error(errorMsg);
184     }
185 
186     // Verify HRegionInfo
187     for (HRegionInfo region : regions) {
188       SnapshotRegionManifest regionManifest = regionManifests.get(region.getEncodedName());
189       if (regionManifest == null) {
190         // could happen due to a move or split race.
191         String mesg = " No snapshot region directory found for region:" + region;
192         if (errorMsg.isEmpty()) errorMsg = mesg;
193         LOG.error(mesg);
194         continue;
195       }
196 
197       verifyRegionInfo(region, regionManifest);
198     }
199 
200     if (!errorMsg.isEmpty()) {
201       throw new CorruptedSnapshotException(errorMsg);
202     }
203 
204     // Verify Snapshot HFiles
205     SnapshotReferenceUtil.verifySnapshot(services.getConfiguration(), fs, manifest);
206   }
207 
208   /**
209    * Verify that the regionInfo is valid
210    * @param region the region to check
211    * @param manifest snapshot manifest to inspect
212    */
213   private void verifyRegionInfo(final HRegionInfo region,
214       final SnapshotRegionManifest manifest) throws IOException {
215     HRegionInfo manifestRegionInfo = HRegionInfo.convert(manifest.getRegionInfo());
216     if (!region.equals(manifestRegionInfo)) {
217       String msg = "Manifest region info " + manifestRegionInfo +
218                    "doesn't match expected region:" + region;
219       throw new CorruptedSnapshotException(msg, snapshot);
220     }
221   }
222 }