[ https://issues.apache.org/jira/browse/JAMES-1635?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15023250#comment-15023250 ]
Tellier Benoit commented on JAMES-1635: --------------------------------------- This evening I had some time to look this issue : - From my understanding, disk IO are blocking (neither RAM nor CPU are when running Cassandra unit tests) >From james-project pom.xml - <forkedProcessTimeoutInSeconds>1800</forkedProcessTimeoutInSeconds> This value is two low for common hardware (always failed on my 5400 turns/minute hard drive and have failures even on SAS hard drives ). I think about increasing this to 3600 in the future. I hope not for long... Even related to disk IO, not yet commited but definitly needed : https://github.com/apache/james-project/pull/17 And finaly, I think we can parallelise Cassandra unit tests : - using different keyspaces. On each invocation on test cluster :: build, we create a new keyspace on the already lauched database (providing thus test isolation) - we then use <parallel>classesAndMethods</parallel> in cassandra related subprojects What do you think of this ? I definitly gonna try this, because, in my opinion, 2h+ long unit tests are not acceptables. I think 20 minutes on commodity hardware should be a maximum... Other workaround : are locales test in MPT really necessary for each feature ? I mean, can't we have a dedicated test, even with stranger encoding trics, but get a rid of this redoundancy ? If anyone have comments / suggestions of any kind... > Can not run unit tests > ---------------------- > > Key: JAMES-1635 > URL: https://issues.apache.org/jira/browse/JAMES-1635 > Project: James Server > Issue Type: Bug > Affects Versions: Trunk > Reporter: Tellier Benoit > Fix For: Trunk > > > I got this exception, either in cassandra mailbox and Cassandra backend ; > For example : > initializeTypesShouldCreateTheTypes(org.apache.james.backends.cassandra.init.CassandraTypeProviderTest) > Time elapsed: 14.353 sec <<< FAILURE! > java.lang.AssertionError: Cassandra daemon did not start within timeout > at > org.cassandraunit.utils.EmbeddedCassandraServerHelper.startEmbeddedCassandra(EmbeddedCassandraServerHelper.java:130) > at > org.cassandraunit.utils.EmbeddedCassandraServerHelper.startEmbeddedCassandra(EmbeddedCassandraServerHelper.java:85) > at > org.cassandraunit.utils.EmbeddedCassandraServerHelper.startEmbeddedCassandra(EmbeddedCassandraServerHelper.java:64) > at > org.cassandraunit.utils.EmbeddedCassandraServerHelper.startEmbeddedCassandra(EmbeddedCassandraServerHelper.java:56) > at > org.cassandraunit.utils.EmbeddedCassandraServerHelper.startEmbeddedCassandra(EmbeddedCassandraServerHelper.java:52) > at > org.apache.james.backends.cassandra.CassandraCluster.<init>(CassandraCluster.java:57) > at > org.apache.james.backends.cassandra.CassandraCluster.create(CassandraCluster.java:51) > at > org.apache.james.backends.cassandra.init.CassandraTypeProviderTest.setUp(CassandraTypeProviderTest.java:66) > I increased the max retry count of the Cassandra cluster but it does not > solve my problem. > It really slow down my worlflow... -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: server-dev-unsubscr...@james.apache.org For additional commands, e-mail: server-dev-h...@james.apache.org