[jira] [Commented] (CASSANDRA-10648) Native protocol is dead after running some Hive queries
[ https://issues.apache.org/jira/browse/CASSANDRA-10648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14992116#comment-14992116 ] Jim Witschey commented on CASSANDRA-10648: -- [~alexliu68] How confident are you that this is a Cassandra issue, rather than a DSE Hive issue? If possible, I'd be interested to see if using a different version of C* on the backend results in the same problem. And, just to clarify, you think C* becomes unavailable because of performance issues, not some other bug making connecting via the default port not work anymore? > Native protocol is dead after running some Hive queries > --- > > Key: CASSANDRA-10648 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10648 > Project: Cassandra > Issue Type: Bug >Reporter: Alex Liu > Fix For: 2.2.x > > Attachments: 10_day_loss.q, deadlock.txt, deadlock1.txt > > > When test on DSE portfolio demo, which basically creates a few C* tables and > inserts some data into the tables, then run some Hive queries on the tables. > I attach the Hive queries > After some queries are done, C* node is dead on native port, cqlsh can't > login any more. > Some thread dumps are attached. Too many threads are in waiting mode and > system is not responding. > It's tested on C* 2.2.3 -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-10648) Native protocol is dead after running some Hive queries
[ https://issues.apache.org/jira/browse/CASSANDRA-10648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14992236#comment-14992236 ] Alex Liu commented on CASSANDRA-10648: -- C* 2.1.x works fine. I shut down all other programs and C* node native is dead forever without any client running on C*. > Native protocol is dead after running some Hive queries > --- > > Key: CASSANDRA-10648 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10648 > Project: Cassandra > Issue Type: Bug >Reporter: Alex Liu > Fix For: 2.2.x > > Attachments: 10_day_loss.q, deadlock.txt, deadlock1.txt > > > When test on DSE portfolio demo, which basically creates a few C* tables and > inserts some data into the tables, then run some Hive queries on the tables. > I attach the Hive queries > After some queries are done, C* node is dead on native port, cqlsh can't > login any more. > Some thread dumps are attached. Too many threads are in waiting mode and > system is not responding. > It's tested on C* 2.2.3 -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (CASSANDRA-10648) Native protocol is dead after running some Hive queries
[ https://issues.apache.org/jira/browse/CASSANDRA-10648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=14988286#comment-14988286 ] Alex Liu commented on CASSANDRA-10648: -- set hive.auto.convert.join=false; in Hive to disable MapJoin > Native protocol is dead after running some Hive queries > --- > > Key: CASSANDRA-10648 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10648 > Project: Cassandra > Issue Type: Bug >Reporter: Alex Liu > Fix For: 2.2.x > > Attachments: 10_day_loss.q, deadlock.txt, deadlock1.txt > > > When test on DSE portfolio demo, which basically creates a few C* tables and > inserts some data into the tables, then run some Hive queries on the tables. > I attach the Hive queries > After some queries are done, C* node is dead on native port, cqlsh can't > login any more. > Some thread dumps are attached. Too many threads are in waiting mode and > system is not responding. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
