wchevreuil commented on a change in pull request #18: HBASE-22567 - HBCK2 
addMissingRegionsToMeta
URL: 
https://github.com/apache/hbase-operator-tools/pull/18#discussion_r320854169
 
 

 ##########
 File path: hbase-hbck2/src/main/java/org/apache/hbase/HBCK2.java
 ##########
 @@ -164,6 +178,102 @@ int setRegionState(String region, RegionState.State 
newState)
     return EXIT_FAILURE;
   }
 
+  Map<TableName,List<Path>> reportTablesWithMissingRegionsInMeta(String... 
nameSpaceOrTable)
+      throws Exception {
+    Map<TableName,List<Path>> report;
+    try(final MetaFixer metaFixer = new MetaFixer(this.conf)){
+      List<String> names = nameSpaceOrTable != null ? 
Arrays.asList(nameSpaceOrTable) : null;
+      report = metaFixer.reportTablesMissingRegions(names);
+    } catch (Exception e) {
+      LOG.error("Error reporting missing regions: ", e);
+      throw e;
+    }
+    if(LOG.isDebugEnabled()) {
+      LOG.debug(formatMissingRegionsInMetaReport(report));
+    }
+    return report;
+  }
+
+  List<String> addMissingRegionsInMeta(List<Path> regionsPath) throws 
IOException {
+    List<String> reAddedRegionsEncodedNames = new ArrayList<>();
+    try(final MetaFixer metaFixer = new MetaFixer(this.conf)){
+      for(Path regionPath : regionsPath){
+        metaFixer.putRegionInfoFromHdfsInMeta(regionPath);
+        reAddedRegionsEncodedNames.add(regionPath.getName());
+      }
+    }
+    return reAddedRegionsEncodedNames;
+  }
+
+  Pair<List<String>, List<ExecutionException>> 
addMissingRegionsInMetaForTables(String...
+      nameSpaceOrTable) {
+    ExecutorService executorService = Executors.newFixedThreadPool(
+      (nameSpaceOrTable == null ||
+        nameSpaceOrTable.length > Runtime.getRuntime().availableProcessors()) ?
+          Runtime.getRuntime().availableProcessors() :
+          nameSpaceOrTable.length);
+    List<Future<List<String>>> futures = new ArrayList<>( nameSpaceOrTable == 
null ? 1 :
+      nameSpaceOrTable.length);
+    final List<String> readdedRegionNames = new ArrayList<>();
+    List<ExecutionException> executionErrors = new ArrayList<>();
+    try {
+      //reducing number of retries in case disable fails due to namespace 
table region also missing
+      this.conf.setInt(HConstants.HBASE_CLIENT_RETRIES_NUMBER, 1);
+      try(Connection conn = ConnectionFactory.createConnection(this.conf);
+        final Admin admin = conn.getAdmin()) {
+        Map<TableName,List<Path>> report = 
reportTablesWithMissingRegionsInMeta(nameSpaceOrTable);
+        for (TableName tableName : report.keySet()) {
+          if(admin.tableExists(tableName)) {
+            futures.add(executorService.submit(new Callable<List<String>>() {
+              @Override
+              public List<String> call() throws Exception {
+                LOG.debug("running thread for {}", 
tableName.getNameWithNamespaceInclAsString());
+                try {
+                  admin.disableTable(tableName);
 
 Review comment:
   My motivation with disabling is trying to minimise potential extra problems 
due to splitting or merging, that may occur in a situation a table is missing 
regions but is still being written to or manipulated by client applications.
   
   > So, if table fails disable, you dump out regions to assign but table could 
be in DISABLING state?
   That's a possibility indeed, and assigns would then fail, but I believe the 
errors printed by assigns would give some clues, and operator could use extra 
hbck2 or hbase shell commands to get it sorted. 
   
   Would you think is wort to remove this extra disable/enable steps?

----------------------------------------------------------------
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


With regards,
Apache Git Services

Reply via email to