Examples of TrackingParallelWriterIndexCommitter

com.salesforce.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter
Like the {@link ParallelWriterIndexCommitter}, but blocks until all writes have attempted to allow the caller to retrieve the failed and succeeded index updates. Therefore, this class will be a lot slower, in the face of failures, when compared to the {@link ParallelWriterIndexCommitter} (though as fast for writes), so it should be used only whenyou need to at least attempt all writes and know their result; for instance, this is fine for doing WAL recovery - it's not a performance intensive situation and we want to limit the the edits we need to retry.
On failure to {@link #write(Multimap)}, we return a {@link MultiIndexWriteFailureException} thatcontains the list of {@link HTableInterfaceReference} that didn't complete successfully.
Failures to write to the index can happen several different ways:
1. this is {@link #stop(String) stopped} or aborted (via the passed {@link Abortable}. This causing any pending tasks to fail whatever they are doing as fast as possible. Any writes that have not begun are not even attempted and marked as failures.
2. A batch write fails. This is the generic HBase write failure - it may occur because the index table is not available, .META. or -ROOT- is unavailable, or any other (of many) possible HBase exceptions.
Regardless of how the write fails, we still wait for all writes to complete before passing the failure back to the client.
org.apache.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter
Like the {@link ParallelWriterIndexCommitter}, but blocks until all writes have attempted to allow the caller to retrieve the failed and succeeded index updates. Therefore, this class will be a lot slower, in the face of failures, when compared to the {@link ParallelWriterIndexCommitter} (though as fast for writes), so it should be used only whenyou need to at least attempt all writes and know their result; for instance, this is fine for doing WAL recovery - it's not a performance intensive situation and we want to limit the the edits we need to retry.
On failure to {@link #write(Multimap)}, we return a {@link MultiIndexWriteFailureException} thatcontains the list of {@link HTableInterfaceReference} that didn't complete successfully.
Failures to write to the index can happen several different ways:
1. this is {@link #stop(String) stopped} or aborted (via the passed {@link Abortable}. This causing any pending tasks to fail whatever they are doing as fast as possible. Any writes that have not begun are not even attempted and marked as failures.
2. A batch write fails. This is the generic HBase write failure - it may occur because the index table is not available, .META. or -ROOT- is unavailable, or any other (of many) possible HBase exceptions.
Regardless of how the write fails, we still wait for all writes to complete before passing the failure back to the client.
org.apache.phoenix.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter
Like the {@link ParallelWriterIndexCommitter}, but blocks until all writes have attempted to allow the caller to retrieve the failed and succeeded index updates. Therefore, this class will be a lot slower, in the face of failures, when compared to the {@link ParallelWriterIndexCommitter} (though as fast for writes), so it should be used only whenyou need to at least attempt all writes and know their result; for instance, this is fine for doing WAL recovery - it's not a performance intensive situation and we want to limit the the edits we need to retry.
On failure to {@link #write(Multimap)}, we return a {@link MultiIndexWriteFailureException} thatcontains the list of {@link HTableInterfaceReference} that didn't complete successfully.
Failures to write to the index can happen several different ways:
1. this is {@link #stop(String) stopped} or aborted (via the passed {@link Abortable}. This causing any pending tasks to fail whatever they are doing as fast as possible. Any writes that have not begun are not even attempted and marked as failures.
2. A batch write fails. This is the generic HBase write failure - it may occur because the index table is not available, .META. or -ROOT- is unavailable, or any other (of many) possible HBase exceptions.
Regardless of how the write fails, we still wait for all writes to complete before passing the failure back to the client.

Examples of com.salesforce.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter

    
        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

Examples of org.apache.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter

    
        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

Examples of org.apache.phoenix.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter


        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

Examples of org.apache.phoenix.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter


        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

Examples of org.apache.phoenix.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter


        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

Examples of org.apache.phoenix.hbase.index.write.recovery.TrackingParallelWriterIndexCommitter


        // setup the actual index writer
        this.writer = new IndexWriter(env, serverName + "-index-writer");
    
        // setup the recovery writer that does retries on the failed edits
        TrackingParallelWriterIndexCommitter recoveryCommmiter =
            new TrackingParallelWriterIndexCommitter();
    
        try {
          // get the specified failure policy. We only ever override it in tests, but we need to do it
          // here
          Class<? extends IndexFailurePolicy> policyClass =
              env.getConfiguration().getClass(INDEX_RECOVERY_FAILURE_POLICY_KEY,
                StoreFailuresInCachePolicy.class, IndexFailurePolicy.class);
          IndexFailurePolicy policy =
              policyClass.getConstructor(PerRegionIndexWriteCache.class).newInstance(failedIndexEdits);
          LOG.debug("Setting up recovery writter with committer: " + recoveryCommmiter.getClass()
              + " and failure policy: " + policy.getClass());
          recoveryWriter =
              new IndexWriter(recoveryCommmiter, policy, env, serverName + "-recovery-writer");
        } catch (Exception ex) {
          throw new IOException("Could not instantiate recovery failure policy!", ex);

View Full Code Here

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.