Closed milandesai closed 9 years ago
Great. Simplifies things. We probably can move GiraffaRpcRetryingCallerFactory
inside org.apache.hadoop.hbase.client
package, and then FileSystemExceptionInterceptor
can be a static sub class inside the factory.
I moved GiraffaRpcRetryingCallerFactory
under hbase.client
and merged everything in one class.
Committed. Thank you, Milan.
Currently, we implement our own makeshift RpcRetryingCaller to avoid retrying filesystem exceptions. With HBase 1.0, a better way is to use an Interceptor that captures Exceptions thrown by the server and rethrows them if they are filesystem-related. This allows non-filesystem related Exceptions to be processed normally and avoids rewriting the RPC retrying implementation.