Stack Overflow | swinefish | 2 weeks ago
  1. 0

    Storm-hbase bolt failing over DRPC

    Stack Overflow | 2 weeks ago | swinefish

    Root Cause Analysis

    1. backtype.storm.generated.DRPCExecutionException

      No message provided

      at backtype.storm.daemon.drpc$service_handler$reify__8688.failRequest()
    2. backtype.storm.daemon
      1. backtype.storm.daemon.drpc$service_handler$reify__8688.failRequest(drpc.clj:136)[storm-core-]
      1 frame
    3. backtype.storm.drpc
      1.[storm-core- A short while later, I get org.apache.hadoop.hbase.client.RetriesExhaustedException. This doesn't always happen, but is very common. My assumption based on this is one of two possibilities: The scan is timing out. However performing the scan through HBase Shell or REST return in less than a second The table is inconsistent, causing a certain region to be missing. I have run hbase hbck and it shows 0 inconsistencies. I know that the connection to HBase is fine: I have added debugging output and the bolt gets the results. However due to the DRPCExecutionException, these results are never returned over DRPC. I though the issue was DRPC timeout, however I have increased the DRPC timeout a lot and I get the same result in the same amount of time. After Googling I found someone else with the same issue ([Storm]
      1 frame