You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Jeff Beck (JIRA)" <ji...@apache.org> on 2016/09/13 13:43:20 UTC

[jira] [Commented] (CASSANDRA-12633) Assert Errors occurring with OutboundTcpConnection causes Cassandra to slowly stop functioning.

    [ https://issues.apache.org/jira/browse/CASSANDRA-12633?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15487236#comment-15487236 ] 

Jeff Beck commented on CASSANDRA-12633:
---------------------------------------

Started at 2.1 but if it isn't critical enough for that I made a patch with it starting at 2.2 as well. Let me know which is preferred and I can create patches for the other branches.

> Assert Errors occurring with OutboundTcpConnection causes Cassandra to slowly stop functioning.
> -----------------------------------------------------------------------------------------------
>
>                 Key: CASSANDRA-12633
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12633
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Jeff Beck
>             Fix For: 2.1.16
>
>         Attachments: 12633-2.1.txt, 12633-2.2-start.txt, 12633-2.2.txt
>
>
> Assert Errors occurring with OutboundTcpConnection causes Cassandra to slowly stop functioning.
> There is a possibility of assert errors in ByteBufferUtil.writeWithShortLength which can cause a stop to processing. There is already good error handling it just needs to catch Throwable instead of Exception  at https://github.com/apache/cassandra/blob/c26bd91852cbf19d7dba9f62078f5da31b04dbe0/src/java/org/apache/cassandra/net/OutboundTcpConnection.java#L239 
> An example of when this happens:
> {code}
> ERROR [MessagingService-Outgoing-/10.0.19.152] 2016-05-03 21:43:54,638 CassandraDaemon.java:227 - Exception in thread Thread[MessagingService-Outgoing-/10.0.19.152,5,main]
> java.lang.AssertionError: 103265
>         at org.apache.cassandra.utils.ByteBufferUtil.writeWithShortLength(ByteBufferUtil.java:290) ~[apache-cassandra-2.1.12.jar:2.1.12]
>    at org.apache.cassandra.db.composites.AbstractCType$Serializer.serialize(AbstractCType.java:392) ~[apache-cassandra-2.1.12.jar:2.1.12]
>      at org.apache.cassandra.db.composites.AbstractCType$Serializer.serialize(AbstractCType.java:381) ~[apache-cassandra-2.1.12.jar:2.1.12]
>      at org.apache.cassandra.db.filter.ColumnSlice$Serializer.serialize(ColumnSlice.java:271) ~[apache-cassandra-2.1.12.jar:2.1.12]
>      at org.apache.cassandra.db.filter.ColumnSlice$Serializer.serialize(ColumnSlice.java:259) ~[apache-cassandra-2.1.12.jar:2.1.12]
>      at org.apache.cassandra.db.filter.SliceQueryFilter$Serializer.serialize(SliceQueryFilter.java:503) ~[apache-cassandra-2.1.12.jar:2.1.12]
>    at org.apache.cassandra.db.filter.SliceQueryFilter$Serializer.serialize(SliceQueryFilter.java:490) ~[apache-cassandra-2.1.12.jar:2.1.12]
>      at org.apache.cassandra.db.SliceFromReadCommandSerializer.serialize(SliceFromReadCommand.java:168) ~[apache-cassandra-2.1.12.jar:2.1.12]
>  at org.apache.cassandra.db.ReadCommandSerializer.serialize(ReadCommand.java:143) ~[apache-cassandra-2.1.12.jar:2.1.12]
>    at org.apache.cassandra.db.ReadCommandSerializer.serialize(ReadCommand.java:132) ~[apache-cassandra-2.1.12.jar:2.1.12]
>    at org.apache.cassandra.net.MessageOut.serialize(MessageOut.java:121) ~[apache-cassandra-2.1.12.jar:2.1.12]
>       at org.apache.cassandra.net.OutboundTcpConnection.writeInternal(OutboundTcpConnection.java:330) ~[apache-cassandra-2.1.12.jar:2.1.12]
>     at org.apache.cassandra.net.OutboundTcpConnection.writeConnected(OutboundTcpConnection.java:282) ~[apache-cassandra-2.1.12.jar:2.1.12]
>    at org.apache.cassandra.net.OutboundTcpConnection.run(OutboundTcpConnection.java:218) ~[apache-cassandra-2.1.12.jar:2.1.12]
> {code}
> The exact error happened when you sent a read request via cql for a clustering column value with more than 64k chars much like CASSANDRA-8280 but on reads. That exact error case is no longer reproducable in the latest versions of C* but in general it seems like it should still handle the possible asserts.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)