[ https://issues.apache.org/jira/browse/HBASE-4652?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Jonathan Hsieh updated HBASE-4652: ---------------------------------- Description: The suggestions from HBASE-4552 and initial implementations could violate row atomicity if there is a failure while bulk-loading multiple CF rows. A mechanism to atomically rollback or commit only upon completion (mvcc at the store level) is needed if a partial failures encountered when attempting to bulk load multiple families. For example, let's say I want to bulk load a region with cfs A, B, C. I issue a call to an RS region to atomically bulkload the HFiles. The RS loads A and B successfully but fails on C (hdfs failure, or rs goes down, etc). We could rollback A and B - if we don't we would have A and B loaded but not C and have an atomicity violation. was: The suggestions from HBASE-4552 and initial implementations could violate row atomicity if there are multiple CFs. A mechanism to atomically rollback or commit only upon completion (mvcc at the store level) is needed if a partial failures encountered when attempting to bulk load multiple families. For example, let's say I want to bulk load a region with cfs A, B, C. I issue a call to an RS region to atomically bulkload the HFiles. The RS loads A and B successfully but fails on C (hdfs failure, or rs goes down, etc). We could rollback A and B - if we don't we would have A and B loaded but not C and have an atomicity violation. > Mechanism for atomic recovery from partial failure when atomically > bulk-loading multi-CF rows. > ---------------------------------------------------------------------------------------------- > > Key: HBASE-4652 > URL: https://issues.apache.org/jira/browse/HBASE-4652 > Project: HBase > Issue Type: Bug > Reporter: Jonathan Hsieh > > The suggestions from HBASE-4552 and initial implementations could violate row > atomicity if there is a failure while bulk-loading multiple CF rows. A > mechanism to atomically rollback or commit only upon completion (mvcc at the > store level) is needed if a partial failures encountered when attempting to > bulk load multiple families. > For example, let's say I want to bulk load a region with cfs A, B, C. I issue > a call to an RS region to atomically bulkload the HFiles. The RS loads A and > B successfully but fails on C (hdfs failure, or rs goes down, etc). We could > rollback A and B - if we don't we would have A and B loaded but not C and > have an atomicity violation. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira