accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christopher Tubbs (JIRA)" <>
Subject [jira] [Commented] (ACCUMULO-4584) Determine need for oneway method handling in RpcWrapper
Date Fri, 10 Feb 2017 21:28:42 GMT


Christopher Tubbs commented on ACCUMULO-4584:

Okay, so this is what I think we should do, based on IRC conversation with [~elserj] and additional
inspection of the code:

# Remove the checks in the wrapper for oneway methods, so we wrap all RuntimeException with
# Rely on thrift 0.10.0 behavior to log and drop oneway exceptions
# Remove wrapping for Error (that's dangerous and unpredictable)
# Remove unnecessary reflection utils to inspect service for oneway methods and simplify wrapper
# Remove unnecessary and broken RpcWrapperTests
# Create simple IDL test case using our RpcWrapper and libthrift to verify expected behavior
for oneway and regular methods (might be separate issue) for regression testing
# If future version of thrift incorporates wrapping of RuntimeException, we delete all our
RpcWrapper stuffs and push regression test case into thrift itself

> Determine need for oneway method handling in RpcWrapper
> -------------------------------------------------------
>                 Key: ACCUMULO-4584
>                 URL:
>             Project: Accumulo
>          Issue Type: Task
>          Components: rpc
>            Reporter: Christopher Tubbs
>            Assignee: Christopher Tubbs
>             Fix For: 2.0.0
> RpcWrapper has checks for oneway methods (from ACCUMULO-4065), but may no longer need
it after upgrading to Thrift 0.10.0, due to the fix in THRIFT-3479
> Should determine whether these are still needed or can be removed.
> The current RpcWrapperTest fails if the checks are removed, but that test code may be
assuming behavior in thrift, since it doesn't seem to actually use thrift code in the test

This message was sent by Atlassian JIRA

View raw message