How Do I Quickly Restore HBase After HDFS Enters the Safe Mode and the HBase Service Is Abnormal?

Symptom

For MRS 3.5.0 and later versions, the HBase service is abnormal after the HDFS enters the safe mode. How do I quickly restore the HBase service after the HDFS exits the safe mode?

Cause Analysis

After the HDFS enters the safe mode, the HBase service is abnormal and the meta table goes offline. After the HDFS exits the safe mode, the meta table does not go online. "No namenode available to invoke create /hbase/WALs/xxxx.meta" is displayed in the RegionServer log.

The meta table cannot record the online status during the online process after the HDFS fault is rectified. In addition, the number of retry times for automatic recovery of the Manager instance health check is limited. As a result, the meta table cannot be brought online. After the HDFS exits the safe mode, you need to manually restore the service.

Procedure

Ensure that the HDFS is restored and exits the safe mode.
Log in to FusionInsight Manager and choose Cluster > Services > HBase. On the Dashboard page, choose More > Restart Service, enter the password of the current user, and click OK to restart the HBase service, and then bring the meta table online.