nodetool sjk

Runs Swiss Java Knife (SJK) commands to execute, troubleshoot, and monitor the database using MBeans.

Synopsis

nodetool sjk options command command_options

Cassandra tarball installations:

installation_location/cassandra/bin

Nodetool general options

The following options apply to all nodetool commands.

Table 1. Options
Short	Long	Description
`-h`	`--host`	Hostname or IP address.
`-p`	`--port`	Port number.
`-pwf`	`--password-file`	Password file path.
`-pw`	`--password`	Password.
`-u`	`--username`	Remote JMX agent username.
`--`	Separates an option from an argument that could be mistaken for a option.

Options

General Swiss Java Knife options.

--commands: Displays a list of all SJK commands.
--help: Displays command specific help, for example nodetool sjk ttop --help.
-X, --verbose: Displays detailed diagnostics.

Commands

DataStax supports running SJK commands from nodetool.

mxdump

Prints serialized MBeans to JSON format.

Warning:

The nodetool sjk mxdump command tries to print all exposed MBeans to the console. The command can fail if it encounters an improperly formatted MBean that exposes a non-serializable type. To interact with a specific list of MBeans, use the nodetool sjk mx command.

Run the nodetool sjk mx command with the -q (query) switch to avoid errors if a non-serializable type (such as an mx4j MBean) is returned.

-q, --query

Output selected MBeans by specifying the ObjectName; match patterns using the asterisk (*) and question mark (?) wildcards.

For example, to get all values for Garbage Collector:

nodetool sjk mxdump -q java.lang:type=GarbageCollector,name=*

Print GC (garbage collection) log messages for remote process.

For example, to print the GC log messages:

nodetool sjk gc

Note: No options for GC.

Heap histogram (hh) prints the class histogram, similar to jmap -histo. When no options are specified, prints complete histogram for all objects.

Note: For package installations, run nodetool with the account the runs the database.

For example, to print a histogram for young with a sample duration of 10000 milliseconds on a system with Cassandra package installation:

sudo -u cassandra nodetool sjk hh --young --sample-depth 10000ms -n 10

-d, --sample-depth

Specify an integer and unit to set the sample duration for --dead-young or --young; default: 10000ms. To set the unit, use the following notation:

ms (milliseconds)
s (seconds)
m (minutes)
h (hours)
d (days)

--dead

Prints histogram of dead objects.

--dead-young

Prints histogram for sample of dead young objects. Use with -d to set the duration.

--live

Prints histogram of live objects.

-n, --top-number

Limits the number of buckets for N top buckets; default: 2147483647.

--young

Prints histrogram for a sample of the new objects. Use with -d to set the duration.

Provides query and execution for Mbeans from the nodetool interface.

-all, --allMatched: Process all matched MBeans.
-a, --arguments: Arguments for MBean operation invocation.
-f, --field, --attribute: MBean attribute
-b, --bean: Mbean name.
-mc, --call: Invokes MBean method. Default false.
-mg, --get: Retrieves value of MBean attribute. Default false.
-mi, --info: Display metadata for MBean. Default false.
--max-col-width: Table column width threshold for formatting tabular data. Default 40.
-op, --operation: MBean operation name to be called.
--quiet: Avoid non-essential output; default: false.
-ms, --set: Sets value for MBean attribute; default: false.
-v, --value: Value to set for MBean attribute.

jps

Enhanced version of jps tool from the Java Development Kit (JDK) that lists the instrumented HotSpot Java Virtual Machines (JVMs) on the target system.

-fd, --filter-description: Use a wildcard expression to match the process description.
-fp, --filter-property: Use a wildcard expressions to match JVM system properties.
-pd, --process-details: Prints custom information related to a process; set to PID, MAIN, FDQN_MAIN, ARGS, Dsys-prop, or Dsys-prop.
Xjvm-flag: Prints custom information related to the specified JVM flag.

stcap

Stack Capture dumps stack traces to file for further processing.

-e, --empty: Set to true to retain threads without stack trace in dump (ignored by default); default: false.
-f, --filter: Filters threads by name using Java RegEx syntax; default: .*
Note: The default value, period (.) followed by asterisk (*), allows any number of any characters, including nulls.
-l, --limit: Target number of traces to collect, once reached command terminates where 0 is unlimited; default: 0.
-m, --match-frame: Set to filtering string and only traces containing are included in the dump.
-o, --output: Write the thread dump to the specified a file name.
-r, --rotate: When specified output file would be rotated every N traces. Specify 0 for no rotation; default: 0.
-i, --sampler-interval: Interval between polling MBeans; default: 0.
-t, --timeout: Time until command terminate even without enough traces collected; default: 30000.

stcpy

Stack Copy utility copies and filters dumps.

-e, --empty: Set to true to retain threads without stack trace in dump (ignored by default); default: false.
-i, --input: Input files.
--mask: One or more masking rules, for example com.mycompany:com.somecompany; default: null.
-m, --match-frame: Frame filter, only traces containing the string are included in dump.
-o, --output: Set to filename. Writes thread dump to the specified file.
-ss, --subsample: If below 1.0 some frames are randomly thrown away. For example, 0.1 retains only every 10th frame; default: 1.0.
-tf, --thread-filter: Filter threads by name using Java RegEx syntax; default: .*

ssa

Stack Sample Analyzer for stack trace dumps.

--categorize: Prints summary for provided categorization; no summary printed by default.
-cf, --categorizer-file: Path to file with stack trace categorization definition.
-co, --csv-output: Output data in CSV format; default: false.
-f, --file: Path to stack dump file.
--flame: Exports flame graph to SVG format; default: not included.
--histo: Prints frame histogram; default: not included.
-nc, --named-class: Defines name stack trace. Use with other options.
classes: Use name=filter expression notation; default: no expression.
--print: Print traces from file; default: not included.
-rc, --rainbow: List of filters for rainbow coloring.
--ssa-help: Provides additional information about SSA options; default: not shown.
-si, --summary-info: Lists summaries.
--thread-info: Prints per thread information summary; default: false.
-tn, --thread-name: Filters thread name using Java RegEx syntax.
-tr, --time-range: Filters time range.
-tz, --time-zone: Time zone used for timestamps; default: UTC. Use --ssa-help for timezone notation.
--title: Inserts a flame graph title; default: Flame Graph.
-tf, --trace-filter: Applies filter to traces before processing. Use --ssa-help for more details about filter notation.
-tt, --trace-trim: Positional filter trim frames to process. Use --ssa-help for more details about trace filters.
--width: Flame graph width in pixels; default: 1200.

ttop

Displays top threads from JVM process.

-f, --filter: Wild card expression to filter threads by name.
-o, --order: Sort order. Value tags: CPU, USER, SYS, ALLOC, NAME; default: CPU.
-ri, --report-interval: Interval between CPU usage reports; default: 10000.
-si, --sampler-interval: Interval between polling MBeans; default: 500.
-n, --top-number: Number of threads to show; default: 20.