[HBASE-16149] Log the underlying RPC exception in RpcRetryingCallerImpl

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 1.2.0
• Fix Version/s: 1.3.0, 2.0.0
• Hadoop Flags: Reviewed
• Assignee: Jerry He
• Reporter: Jerry He
• Created: 30/Jun/16 05:28
• Updated: 09/Nov/17 01:13
• Resolved: 30/Jun/16 12:59

Description

In RpcRetryingCallerImpl:

  public T callWithRetries(RetryingCallable<T> callable, int callTimeout)
  throws IOException, RuntimeException {
...
    for (int tries = 0;; tries++) {
      try {
...
        return callable.call(getTimeout(callTimeout));
...
      } catch (Throwable t) {
        ExceptionUtil.rethrowIfInterrupt(t);
        if (tries > startLogErrorsCnt) {
          LOG.info("Call exception, tries=" + tries + ", maxAttempts=" + maxAttempts + ", started="
              + (EnvironmentEdgeManager.currentTime() - tracker.getStartTime()) + " ms ago, "
              + "cancelled=" + cancelled.get() + ", msg="
              + callable.getExceptionMessageAdditionalDetail());
        }
...

We log the callable.getExceptionMessageAdditionalDetail() msg. But callable.getExceptionMessageAdditionalDetail() may not provide the underlying cause..

For example, in AbstractRegionServerCallable,

  public String getExceptionMessageAdditionalDetail() {
    return "row '" + Bytes.toString(row) + "' on table '" + tableName + "' at " + location;
  }

Let's add the underlying exception cause to the message as well.