[
https://issues.apache.org/jira/browse/CASSANDRA-17026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17426368#comment-17426368
]
Caleb Rackliffe edited comment on CASSANDRA-17026 at 10/8/21, 8:49 PM:
-----------------------------------------------------------------------
To summarize, when protocol version decoding fails in the pre-V5 message
handling code, we don't consume/skip the remaining {{Envelope}} bytes. So, we
log on the initial decoding attempt, but then log again when the channel
closure logic tries to decode again, which leads to all sorts of possible racy
behavior in the test, which assumes a single log message per request.
PR: https://github.com/apache/cassandra/pull/1257
CircleCI:
https://app.circleci.com/pipelines/github/maedhroz/cassandra?branch=CASSANDRA-17026
was (Author: maedhroz):
PR: https://github.com/apache/cassandra/pull/1257
CircleCI:
https://app.circleci.com/pipelines/github/maedhroz/cassandra?branch=CASSANDRA-17026
> UnableToParseClientMessageFromBlockedSubnetTest.badMessageCausesProtocolExceptionFromExcludeList
> failing on extra log entries
> -----------------------------------------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-17026
> URL: https://issues.apache.org/jira/browse/CASSANDRA-17026
> Project: Cassandra
> Issue Type: Bug
> Components: Observability/Logging, Test/dtest/java
> Reporter: Caleb Rackliffe
> Assignee: Caleb Rackliffe
> Priority: Normal
> Fix For: 4.1
>
>
> This test, while parameterized and run more than once, expects a single log
> entry describing the reporting exclusion for the single query made with each
> parameter combination.
> However, we’re seeing this:
> {noformat}
> junit.framework.AssertionFailedError:
> Expected size:<1> but was:<2> in:
> <["DEBUG [nioEventLoopGroup-5-2] {instance:id} 2021-10-07T09:49:42,974
> Excluding client exception for /127.0.0.1:56628; address contained in
> client_error_reporting_exclusions",
> "DEBUG [nioEventLoopGroup-5-3] {instance:id} 2021-10-07T09:49:42,984
> Excluding client exception for /127.0.0.1:56634; address contained in
> client_error_reporting_exclusions"]>
> at
> org.apache.cassandra.distributed.test.UnableToParseClientMessageFromBlockedSubnetTest.badMessageCausesProtocolExceptionFromExcludeList(UnableToParseClientMessageFromBlockedSubnetTest.java:114)
> at
> java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
> at
> java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
> at
> java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> {noformat}
> Looking at the logs more deeply, the V3 and V4 protocol cases are actually
> logging twice, once on initially processing the message, and again on channel
> closure. There should be a way to eliminate the double logging...
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
