tsquery Language

A metric expression generates the time series. It is a comma-delimited list of one or more metric expression statements. A metric expression statement is the name of a metric, a metric expression function, or a scalar value, joined by one or more metric expression operators.

See the FAQ which answers questions concerning how to discover metrics and use cases for scalar values. For a list of all the supported metrics, see Cloudera Manager Metrics.

Metric expressions support the binary operators: +, -, *, /.

select dt0(total_cpu_user)/getHostFact(numCores,1)*100,
dt0(total_cpu_system)/getHostFact(numCores,1)*100 
where category=ROLE and clusterId=1

You use the AND and OR operators to compose compound predicates.

SELECT fd_open WHERE roleType=DATANODE and active=false

select total_cpu_user where category=HOST

select jvm_committed_heap_mb where category=ROLE and hostname="host1"

How do I compare information across hosts?

1. Click Hosts in the top navigation bar and click a host link.
2. In the Charts pane, choose a chart, for example Host CPU Usage and select and then Open in Chart Builder.
3. In the text box, remove the where entityName=$HOSTID clause and click Build Chart.
4. In the Facets list, click hostname to compare the values across hosts.
5. Configure the time scale, minimums and maximums, and dimension. For example:
   
   
   
   

How do I compare all disk IO for all the DataNodes that belong to a specific HDFS service?

Use a query of the form:

select bytes_read, bytes_written where roleType=DATANODE and serviceName=hdfs1

replacing hdfs1 with your HDFS service name. Then facet by metricDisplayName and compare all DataNode byte_reads and byte_writes metrics at once. See Grouping (Faceting) Time Series for more details about faceting.

When would I use a derivative function?

Some metrics represent a counter, for example, bytes_read. For such metrics it is sometimes useful to see the rate of change instead of the absolute counter value. Use dt or dt0 derivative functions.

When should I use the dt0 function?

Some metrics, like bytes_read represent a counter that always grows. For such metrics a negative rate means that the counter has been reset (for example, process restarted, host restarted, and so on). Use dt0 for these metrics.

How do I display a threshold on a chart?

Suppose that you want to retrieve the latencies for all disks on your hosts, compare them, and show a threshold on the chart to easily detect outliers. Use the following query to retrieve the metrics and the threshold:

select await_time, await_read_time, await_write_time, 250 where category=disk

Then choose All Combined (1) in the Facets list. The scalar threshold 250 will also be rendered on the chart:

See Grouping (Faceting) Time Series for more details about faceting.

I get the warning "The query hit the maximum results limit". How do I work around the limit?

There is a limit on the number of results that can be returned by a query. When a query results in more time-series streams than the limit a warning for "partial results" is issued. To circumvent the problem, reduce the number of metrics you are trying to retrieve or see Configuring Time-Series Query Results.

You can use the rlike operator to limit the query to a subset of entities. For example, instead of

select await_time, await_read_time, await_write_time, 250 where category=DISK

you can use

select await_time, await_read_time, await_write_time, 250 where category=DISK and hostname rlike "host1[0-9]?.cloudera.com"

The latter query retrieves the disk metrics for ten hosts.

How do I discover which metrics are available for which entities?

• Type Select in the text box and then press Space or continue typing. Metrics matching the letters you type display in a drop-down list.
• Select Charts > Chart Builder, click the question mark icon to the right of the Build Chart button and click the List of Metrics link
• Retrieve all metrics for the type of entity:

  select * where roleType=DATANODE