Reprint: http://ju.outofmemory.cn/entry/50064
With the drive of Big data table application, our hbase cluster is getting bigger and larger, however, due to some uncertainties in machine, network and HBase, the system is faced with some uncertain faults.
Therefore, HBase has a lot of region components and needs to control the state of each table's region. Analysis: 1) control the status of region in real time. Each access to an app is directly associated with a region of hbase, and you need to detect if the region on the table is available. 2) The read and write of region is correlated with the state of the underlying HDFS. This correlation determines the monitoring of the read and write status of the region, and also reflects the status of HDFs. 2, the actual combat tool org.apache.hadoop.hbase.tool.Canary monitoring the availability and read and write status of region. ==> the first two problems in correspondence analysis. How to use: Usage:bin/hbase org.apache.hadoop.hbase.tool.Canary [opts]
Where [opts] is:
-help Show this Help and exit.
-daemon continuous check at defined intervals.
-interval <N> Interval between checks (sec) execution ${hbase_home}/bin/hbase Org.apache.hadoop.hbase.tool.Canary detects all the table in the cluster under different configurations, the results are as follows: It will remove the startkey of region by default, Perform a get operation separately according to Columnfamily and print out the system delay. In the case of a problem in region, the state of failed is printed out.
HBase Summary: How to monitor the performance of region