cassandra-stress user

Interleave user provided queries with configurable ratio and distribution.

Synopsis

cassandra-stress user [arguments]

Command options

cl=?: Set the consistency level to use during cassandra-stress. Options are ONE, QUORUM, LOCAL_ONE (default), LOCAL_QUORUM, EACH_QUORUM, ALL, and ANY.
clustering=DIST(?): Distribution clustering runs of operations of the same kind.
duration=?: Specify the time to run, in seconds, minutes, or hours.
err<?: Specify a standard error of the mean. When this value is reached, cassandra-stress will end.

Default: 0.02
n>?: Specify a minimum number of iterations to run before accepting uncertainly convergence.
n<?: Specify a maximum number of iterations to run before accepting uncertainly convergence.
n=?: Specify the number of operations to run.
no-warmup: Do not warmup the process. Do a cold start.
ops(?): Specify what operations to run and the number of each. Only valid with the user option.
profile=?: Designate the YAML file to use with cassandra-stress. Only valid with the user option.
truncate=?: Truncate the table created during cassandra-stress. Options are never, once, always.

Default: never

The truncate option must be inserted before the mode argument, otherwise the cassandra-stress tool won’t apply truncation as specified.

Command arguments

-col

Column details, such as size and count distribution, data generator, names, and comparator:

Supports multiple syntax formats:

-col names=? [slice] [super=?] [comparator=?] [timestamp=?] [size=DIST(?)]

-col [n=DIST(?)] [slice] [super=?] [comparator=?] [timestamp=?] [size=DIST(?)]

-errors

How to handle errors when encountered during stress testing:

-errors [retries=N] [ignore] [skip-read-validation]

retries=<N>: Number of times to try each operation before failing.
ignore: If included, won’t fail on errors. Omit this argument to fail on errors.
skip-read-validation: Include to skip read validation and message output.

-graph

Graph results of cassandra-stress tests. Multiple tests can be graphed together.

-graph file=? [revision=?] [title=?] [op=?]

-insert

Insert specific options relating to various methods for batching and splitting partition updates.

-insert [revisit=DIST(?)] [visits=DIST(?)] partitions=DIST(?) [batchtype=?] select-ratio=DIST(?) row-population-ratio=DIST(?)

-log

Where to log progress and the interval to use.

-log [level=?] [no-summary] [file=?] [hdrfile=?] [interval=?] [no-settings] [no-progress] [show-queries] [query-log-file=?]

-mode

Thrift or CQL with options. Supports multiple syntax formats:

-mode thrift [smart] [user=?] [password=?]

-mode native [unprepared] cql3 [compression=?] [port=?] [user=?] [password=?] [auth-provider=?] [maxPending=?] [connectionsPerHost=?] [protocolVersion=?]

-mode simplenative [prepared] cql3 [port=?]

-node

Nodes to connect to.

-node [datacenter=?] [whitelist] [file=?]

-pop

Population distribution and intra-partition visit order. Supports multiple syntax formats:

-pop seq=? [no-wrap] [read-lookback=DIST(?)] [contents=?]

-pop [dist=DIST(?)] [contents=?]

-port

Specify port for connecting nodes. Port can be specified for Apache Cassandra native protocol (native=), Thrift protocol (thrift=), or a JMX port (jmx=) for retrieving statistics.

-port [native=?] [thrift=?] [jmx=?]

-rate

Set the rate. Supports multiple syntax formats:

Throttle to a fixed rate of operations per second with a specified number of threads:
```
-rate threads=? [throttle=?] [fixed=?]
```
- threads=?: Number of clients to run concurrently.
- throttle=?: Throttle operations per second across all clients to a maximum rate (or less) with no implied schedule. Default: 0
- fixed=?: Expect fixed rate of operations per second across all clients with implied schedule. Default: 0
Gradually increase the number of threads until a specified minimum or maximum is reached, or until throughput saturates:
```
-rate [threads>=?] [threads<=?] [auto]
```
- threads>=?: Run at least this many clients concurrently. Default: 4
- threads⇐?: Run at most this many clients concurrently. Default: 1000
- auto: Stop increasing threads once throughput saturates.

-schema

Schema configuration including replication settings, keyspace, compression, and compaction:

-schema [replication(?)] [keyspace=?] [compaction(?)] [compression=?]

-sendto

Specify a server to send the stress command to.

-sendto <host>

-tokenrange

Token range settings.

-tokenrange [no-wrap] [split-factor=?] [savedata=?]

-transport

Custom transport factories.

-transport [factory=?] [truststore=?] [truststore-password=?] [keystore=?] [keystore-password=?] [ssl-protocol=?] [ssl-alg=?] [store-type=?] [ssl-ciphers=?]

In Apache Cassandra 3.2 and later, the -graph option provides visual feedback for cassandra-stress tests. A file must be named to build the resulting HTML file. A title and revision are optional, but revision must be used if multiple stress tests are graphed on the same output.

cassandra-stress user profile=tools/cqlstress-example.yaml ops\(insert=1\) -graph file=test.html title=test revision=test1

An interactive graph can be displayed with a web browser:

Use a YAML file to run cassandra-stress

This example uses a YAML file named cqlstress-example.yaml, which contains the keyspace and table definitions, and a query definition. The keyspace name and definition are the first entries in the YAML file:

keyspace: perftesting

keyspace_definition:

  CREATE KEYSPACE perftesting WITH replication = { 'class': 'SimpleStrategy', 'replication_factor': 3};

The table name and definition are created in the next section using CQL:

table: users

table_definition:

  CREATE TABLE users (
    username text,
    first_name text,
    last_name text,
    password text,
    email text,
    last_access timeuuid,
    PRIMARY KEY(username)
  );

In the extra_definitions section you can add secondary indexes or materialized views to the table:

extra_definitions:
  - CREATE MATERIALIZED VIEW perftesting.users_by_first_name AS SELECT * FROM perftesting.users WHERE first_name IS NOT NULL and username IS NOT NULL PRIMARY KEY (first_name, username);
  - CREATE MATERIALIZED VIEW perftesting.users_by_first_name2 AS SELECT * FROM perftesting.users WHERE first_name IS NOT NULL and username IS NOT NULL PRIMARY KEY (first_name, username);
  - CREATE MATERIALIZED VIEW perftesting.users_by_first_name3 AS SELECT * FROM perftesting.users WHERE first_name IS NOT NULL and username IS NOT NULL PRIMARY KEY (first_name, username);

The population distribution can be defined for any column in the table. This section specifies a uniform distribution between 10 and 30 characters for username values in gnerated rows, that the values in the generated rows willcreates , a uniform distribution between 20 and 40 characters for generated startdate over the entire Apache Cassandra cluster, and a Gaussian distribution between 100 and 500 characters for description values.

columnspec:
  - name: username
    size: uniform(10..30)
  - name: first_name
    size: fixed(16)
  - name: last_name
    size: uniform(1..32)
  - name: password
    size: fixed(80) # sha-512
  - name: email
    size: uniform(16..50)
  - name: startdate
    cluster: uniform(20...40)
  - name: description
    size: gaussian(100...500)

After the column specifications, you can add specifications for how each batch runs. In the following code, the partitions value directs the test to use the column definitions above to insert a fixed number of rows in the partition in each batch:

insert:
  partitions: fixed(10)
  batchtype: UNLOGGED

The last section contains a query, read1, that can be run against the defined table.

queries:
  read1:
    cql: select * from users where username = ? and startdate = ?
    fields: samerow     # samerow or multirow (select arguments from the same row, or randomly from all rows in the partition)

The following example shows using the user option and its parameters to run cassandra-stress tests from cqlstress-example.yaml:

cassandra-stress user profile=tools/cqlstress-example.yaml n=1000000 ops\(insert=3,read1=1\) no-warmup cl=QUORUM

Notice that:

The user option is required for the profile and opt parameters.
The value for the profile parameter is the path and filename of the .yaml file.
In this example, -n specifies the number of batches that run.
The values supplied for ops specifies which operations run and how many of each. hese values direct the command to insert rows into the database and run the read1 query.

How many times? Each insert or query counts as one batch, and the values in ops determine how many of each type are run. Since the total number of batches is 1,000,000, and ops says to run three inserts for each query, the result will be 750,000 inserts and 250,000 of the read1 query.

Use escaping backslashes when specifying the ops value.

cassandra-stress user

Synopsis

Command options

Command arguments

Use the -graph option

Use a YAML file to run cassandra-stress

Was this helpful?

Give Feedback