Alexey Serbin has posted comments on this change. ( http://gerrit.cloudera.org:8080/12647 )
Change subject: [TS heartbeater] avoid reconnecting to master too often ...................................................................... Patch Set 1: (2 comments) http://gerrit.cloudera.org:8080/#/c/12647/1//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/12647/1//COMMIT_MSG@11 PS1, Line 11: is accepting connections and Ping RPC : requests but isn't able to properly respond to TS heartbeats > Looking at the code, why does this situation return a NetworkError? Shouldn This situation didn't return a NetworkError. It returned ServiceUnavailable error. http://gerrit.cloudera.org:8080/#/c/12647/1/src/kudu/tserver/heartbeater.cc File src/kudu/tserver/heartbeater.cc: http://gerrit.cloudera.org:8080/#/c/12647/1/src/kudu/tserver/heartbeater.cc@595 PS1, Line 595: * At least once per (heartbeat_rpc_timeout ^ 2 / heartbeat_interval) : // time interval; the worst case is when every HB request times out. > Where does this factor in? What's the idea behind this formula? I don't get The most appropriate thing I thought was getting rid of that condition at all and leaving the Network error only. However, I'm a bit concerned that in some bad cases it will be a timeout (not manifested in NetworkError as is) working with a proxy due to a network problem, and re-creating a proxy might help to overcome this situation. Basically, I would opt for having something like 'consecutive errors for more than X seconds' as a second criterion to reset the master proxy. -- To view, visit http://gerrit.cloudera.org:8080/12647 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: kudu Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I961ae453ffd6ce343574ce58cb0e13fdad218078 Gerrit-Change-Number: 12647 Gerrit-PatchSet: 1 Gerrit-Owner: Alexey Serbin <[email protected]> Gerrit-Reviewer: Alexey Serbin <[email protected]> Gerrit-Reviewer: Andrew Wong <[email protected]> Gerrit-Reviewer: Kudu Jenkins (120) Gerrit-Reviewer: Will Berkeley <[email protected]> Gerrit-Comment-Date: Tue, 05 Mar 2019 21:32:34 +0000 Gerrit-HasComments: Yes
