[jira] [Commented] (CASSANDRA-12740) cqlsh copy tests hang in case of no answer from the server or driver

["Stefania (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CASSANDRA-12740?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15547815#comment-15547815 ] 

Stefania commented on CASSANDRA-12740:
--------------------------------------

Here is the patch:

||3.0||3.X||trunk||
|[patch|https://github.com/stef1927/cassandra/commits/12740-cqlsh-3.0]|[patch|https://github.com/stef1927/cassandra/commits/12740-cqlsh-3.X]|[patch|https://github.com/stef1927/cassandra/commits/12740-cqlsh]|
|[cqlsh tests|http://cassci.datastax.com/view/Dev/view/stef1927/job/stef1927-12740-cqlsh-3.0-cqlsh-tests/]|[cqlsh tests|http://cassci.datastax.com/view/Dev/view/stef1927/job/stef1927-12740-cqlsh-3.X-cqlsh-tests/]|[cqlsh tests|http://cassci.datastax.com/view/Dev/view/stef1927/job/stef1927-12740-cqlsh-cqlsh-tests/]|

The 3.0 cqlsh tests are running, I will start the tests on the other branches and move the ticket to patch available when the 3.0 tests are clean.

I noticed that if we set a request timeout, we receive operation-timed-out errors for rows that were already acknowledged, even with the latest driver version (cassandra_test HEAD). Is this a known issue or shall I create a ticket [~aholmber]? 

> cqlsh copy tests hang in case of no answer from the server or driver
> --------------------------------------------------------------------
>
>                 Key: CASSANDRA-12740
>                 URL: https://issues.apache.org/jira/browse/CASSANDRA-12740
>             Project: Cassandra
>          Issue Type: Bug
>            Reporter: Stefania
>            Assignee: Stefania
>
> -If we bundle the driver to cqlsh using the 3.6.0 tag or cassandra_test head, some cqlsh copy tests hang, for example {{test_bulk_round_trip_blogposts}}. See CASSANDRA-12736 and CASSANDRA-11534 for some sample failures.-
> If the driver fails to invoke a callback (either error or success), or if the server never answers to the driver, then the copy parent process will wait forever to receive an answer from child processes. We should put a cap to this. We should also use a very high timeout rather than None, so that the driver will notify us if there is no answer from the server.



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