[ https://issues.apache.org/jira/browse/CASSANDRA-5542?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Yuki Morishita updated CASSANDRA-5542: -------------------------------------- Attachment: 0001-change-streaming-procedure.patch Thanks for the review. Updated 0001 patch. bq. Does separating initiated and receiving stream in StreamManager really buys us anything? (outside slightly complicating things) We need to keep StreamResultFuture of the same plan ID when we are doing streaming on the same JVM. This happens when we bulk load via JMX method, or running StreamingTransferTest unit test. bq. On failure (StreamSession.onError), we could (and probably should) send a message as long as the outgoing connection is opened (even if the incoming one is broken), so the current isConnected() is possibly a bit restrictive. You are right. I changed to put null check on outgoing message handler before sending message instead. > BulkLoader is broken in trunk > ----------------------------- > > Key: CASSANDRA-5542 > URL: https://issues.apache.org/jira/browse/CASSANDRA-5542 > Project: Cassandra > Issue Type: Bug > Affects Versions: 2.0 beta 1 > Reporter: Yuki Morishita > Assignee: Yuki Morishita > Fix For: 2.0 beta 2 > > Attachments: 0001-change-streaming-procedure.patch, > 0002-make-BulkLoader-work-with-SSTableReader.patch, > 0003-update-for-post-CASSANDRA-5171.patch, 5542-fix-NPE.txt, 5542.txt > > > After CASSANDRA-5015 and CASSANDRA-5521, we need CFMetaData to open > SSTable(Reader), especially to get bloom_filter_fp_chance and index_interval. > When using bulkloader, CFMetaData is not available, so we cannot open SSTable > to be streamed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira