[
https://issues.apache.org/jira/browse/CASSANDRA-5481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14078125#comment-14078125
]
Philip Thompson commented on CASSANDRA-5481:
--------------------------------------------
The issue reproduces on 1.2-HEAD and 2.0-HEAD, but not on 2.1. The patch still
applies relatively cleanly on 1.2 and 2.0, and fixes the issue on both.
> CQLSH exception handling could leave a session in a bad state
> -------------------------------------------------------------
>
> Key: CASSANDRA-5481
> URL: https://issues.apache.org/jira/browse/CASSANDRA-5481
> Project: Cassandra
> Issue Type: Bug
> Affects Versions: 1.2.4
> Environment: cqlsh 2.3.0 | Cassandra 1.2.4 | CQL spec 3.0.0 | Thrift
> protocol 19.35.0
> Reporter: Jordan Pittier
> Assignee: Philip Thompson
> Priority: Minor
> Fix For: 1.2.19, 2.0.10
>
> Attachments: 5481.diff, CQLSession.png
>
>
> Playing with CTRL+C in a cqlsh session can leave the (Thrift|Native)
> connection in a bad state.
> To reproduce :
> 1) Run a long running COPY FROM command (COPY test (k, v) FROM
> '/tmp/test.csv')
> 2) Interrupt the importer with CTRL+C
> Repeat step 1 and 2 until you start seeing weird things in the cql shell (see
> attached screenshot)
> The reason is, I believe, the connection (and the cursor) is not correclty
> closed and reopened on interruption.
> I am working to propose a fix.
> Jordan
--
This message was sent by Atlassian JIRA
(v6.2#6252)
