NameNode Heap Usage (Daily) Alert

[#1]

Last night I loaded some heavy amount of data to my hbase cluster . When I started my cluster today morning, it started throwing the above alert.

I tried to increase: namenode_opt_newsize, namenode_opt_maxnewsize, hbase_master_heapsize, hbase_master_xmn_size, metrics_collector_heapsize in AMS, but no luck.

Even I tried to 'expunge' the trash in namenode as well, but the alert still persists.

How can I get rid of this alert?

[#2 · Accepted solution]

@rudra prasad biswas

Increasing or decreasing the heap for those components will not help here. As NameNode Heap Usage (Daily) Alert is a watcher alert.

It watches growth rates of HDFS (in a days time) and if the threshold defined in the alert is reached then it triggers the warning/critical alert message. You should be able to edit the parameters values for this alert in

Ambari from Alerts (Tab) --> search for filter text 'NameNode Heap Usage' on the the 'Alert Definition Name' box.

Edit the alert threshold here.

[#3]

The definition of the alert says "This service-level alert is triggered if the NameNode heap usage deviation has grown beyond the specified threshold within a day period." Initially I had configured "Growth Rate" 50% (CRIT) and 20% (WARN). Now I increased the CRIT value to 60%. Is that the right way to resolve it or I need to try somewhere else?

[#4]

That is the correct place to change.

[#5]

大佬，我也遇到了这个问题，请问这个告警怎么消除呢？