Patrick created HBASE-6343:
------------------------------
Summary: After migrating from one cluster to another, new hmaster
dies when reaching out to old nodes
Key: HBASE-6343
URL: https://issues.apache.org/jira/browse/HBASE-6343
Project: HBase
Issue Type: Bug
Components: migration
Affects Versions: 0.90.6
Environment: centos 6.0
Reporter: Patrick
I am migrating from cluster (movetest-a.domain.com to another
(movetest-b.domain.com), both at the same version of hdfs/hbase. I am using
mozilla's Backup tool to copy the /hbase files, but I imagine any copy tool
would have this issue. Each cluster has its own zookeeper & HDFS.
After copying the files, I try to start the new hmaster (movetest-b), and it
dies because it cannot resolve the data nodes from the source cluster (eg
data01.movetest-a.domain.com).
Log: http://pastie.org/4212244
Things I tried that did not work:
- delete -ROOT- from the copied data
- delete .META. from the copied data
- cleanly shutdown the source before copying (to flush to disk)
Things I tried that did work:
- add the source nodes to /etc/hosts, pointing at 127.0.0.2 (which give
connection refused):
127.0.0.2 data01.movetest-a.domain.com data02.movetest-a.domain.com
data03.movetest-a.domain.com
After bringing up the new cluster (with /etc/hosts in place), I was able to
delete the /etc/hosts entries, and bounce hbase without error.
--
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
