[ https://issues.apache.org/jira/browse/CASSANDRA-10130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15989138#comment-15989138 ]
Andrés de la Peña commented on CASSANDRA-10130: ----------------------------------------------- The most straightforward solution would be to use the write path, as it is done with materialized views and CDC. As we want to avoid this, we could use [a new table|https://github.com/adelapena/cassandra/blob/10130-trunk/src/java/org/apache/cassandra/db/SystemKeyspace.java#L147] in the system keyspace to keep track of indexes that should be rebuilt before start. So, we could mark the indexes to be rebuilt [right before|https://github.com/adelapena/cassandra/blob/10130-trunk/src/java/org/apache/cassandra/streaming/StreamReceiveTask.java#L231] adding the sstables to the column family and rebuilding the index. If everything is ok, we [unmark|https://github.com/adelapena/cassandra/blob/10130-trunk/src/java/org/apache/cassandra/streaming/StreamReceiveTask.java#L238] the indexes for rebuilding. Otherwise, if the node dies, the next node start would [detect|https://github.com/adelapena/cassandra/blob/10130-trunk/src/java/org/apache/cassandra/index/internal/CassandraIndex.java#L173] the indexes as marked for rebuilding and would throw a full index rebuild. [Here|https://github.com/adelapena/cassandra/commit/cd5c3faab5d59cc75b98a8151f6245034c634e53] is the draft of the approach. > Node failure during 2i update after streaming can have incomplete 2i when > restarted > ----------------------------------------------------------------------------------- > > Key: CASSANDRA-10130 > URL: https://issues.apache.org/jira/browse/CASSANDRA-10130 > Project: Cassandra > Issue Type: Bug > Components: Coordination > Reporter: Yuki Morishita > Assignee: Andrés de la Peña > Priority: Minor > > Since MV/2i update happens after SSTables are received, node failure during > MV/2i update can leave received SSTables live when restarted while MV/2i are > partially up to date. > We can add some kind of tracking mechanism to automatically rebuild at the > startup, or at least warn user when the node restarts. -- This message was sent by Atlassian JIRA (v6.3.15#6346) --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@cassandra.apache.org For additional commands, e-mail: commits-h...@cassandra.apache.org