Unable to restart HBase when HDFS is shut down before HBase

Bug #1339454 reported by Hans Zeller
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Trafodion
Fix Released
Undecided
Hans Zeller

Bug Description

We found this problem when our swstopall script shut down HBase after shutting down HDFS, while storing HBase data in HDFS. When we did this, the HMaster process was unable to complete initialization when HDFS/HBase were restarted.

Method to reproduce:

- Use a recent build (with install_local_hadoop using HDFS storage for HBase)
- Modify the swstopall script to call swstophbase after swstophadoop
- Run install_local_hadoop
- Do sqstart, sqlci initialize trafodion
- sqstop
- swstopall
- swstartall
- hbase shell status 'detailed'

The status 'detailed' command will not list all the metadata tables. Also, the HBase output file will not have the "org.apache.hadoop.hbase.master.HMaster: Master has completed initialization" message. The HMaster will die after a while.

Tags: dtm hang
Revision history for this message
Hans Zeller (hans-zeller) wrote :

Fixed by changing the sequence of start/stop operations in the swstartall and swstopall shell scripts.

Changed in trafodion:
status: New → Fix Released
assignee: nobody → Hans Zeller (hans-zeller)
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.