1 In HTable, allow setting an operation timeout on meta. Needed testing.
3 In ServerCallable, do a little refactor so cleaner around checking for timeout.
5 In RpcClient, add back the getRpcTimeout that takes a default timeout and
6 returns minimum of default rpc timeout and whatever is set in the thread local.
7 Call this new method on setup of a 'Service' the first time (In 0.94 it was
8 called when we set up a new proxy, which equates to about the same thing).
10 In TestClientNoCluster, add a test that verifies we get SocketTimeoutException
11 when the operation timeout is reached rather than a retries exhausted
12 exception which is what we were getting when before this patch.
14 Add a log4j.properties to hbase-client so we get a bit of logging.