Andrew Wong has uploaded this change for review. ( http://gerrit.cloudera.org:8080/12002
Change subject: client: unify leader master retry logic ...................................................................... client: unify leader master retry logic In preparation for implementing an asynchronous RPC that will need to connect to the leader master, I've been looking through some of the retry logic we have scattered around the C++ client. I noticed near-identical code in a couple places (LookupRpc and SyncLeaderMasterRpc), so I've ripped it out and put it in its own reusable method. The behavior is mostly the same as before with a few exceptions: - RPC errors in that result in OK Statuses are evaluated up-front and a retry decision is based off of the error. It wasn't clear why this wasn't always the case before, since presumably we always want to handle RPC errors. - GetTableLocations, on TimedOut error, will now retry for the single-master case. - GetTableLocations, on ServiceUnavailable error, will not attempt to reconnect to the leader master, and instead, will simply retry. Change-Id: I2450676da1c723a247c84deb1b895f116173670e --- M src/kudu/client/client-internal.cc M src/kudu/client/client-internal.h M src/kudu/client/client-test.cc M src/kudu/client/meta_cache.cc 4 files changed, 146 insertions(+), 118 deletions(-) git pull ssh://gerrit.cloudera.org:29418/kudu refs/changes/02/12002/1 -- To view, visit http://gerrit.cloudera.org:8080/12002 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: newchange Gerrit-Change-Id: I2450676da1c723a247c84deb1b895f116173670e Gerrit-Change-Number: 12002 Gerrit-PatchSet: 1 Gerrit-Owner: Andrew Wong <aw...@cloudera.com>
