Steve Loughran

> Do you want to keep providing scope optional or should we wrap this up under a profile? 1. the scope in this PR doesn't actually change protobuf 2.5 being...

updated pr tries to address reviews, including building.txt details. regarding @apurtell's comment about static vs qualified import of `getRemoteException()` I started to convert, but got fed up with the needless...

@ayushtkn thanks. i will do it everywhere; its the same as the s3a invoker stuff and it is more elegant, especially for anyone writing new rpc stuff

rebased for a retest, then will merge.

test failure seems unrelated. style ``` ./hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/ipc/protocolPB/GenericRefreshProtocolClientSideTranslatorPB.java:62: .setIdentifier(identifier): '.' has incorrect indentation level 6, expected level should be 8. [Indentation] ./hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/ipc/protocolPB/GenericRefreshProtocolClientSideTranslatorPB.java:63: .addAllArgs(argList): '.' has incorrect indentation level 6, expected level...

dfs upgrade test is flaky (and has a history), created HDFS-17224 to cover the new failure and verified it works standalone

@xizhu-mstr @tooptoop4 i'm not actively working on this; too many pressing issues and after getting 3.3.5 out the door I'm catching up with the internal stuff. Either of you two...

usual ongoing protobuf issues; AFAIK none of them lethal. YMMV I do want #4996 in so we can get protobuf 2.5 off the classpath. if you could take that up,...

says 3.21.x... we should take the latest one we can which doesn't include other surprises...pr and jira can be set to the final version which goes in as it is...

> I see AbstractFSContract but that seems pretty bare, im not sure if thats what you are referring to or something else. look at all the subclasses of `AbstractFSContractTestBase` ......