A collection of command line tools for Crate developers (and maybe users as well).

• Why cr8? 🤔
• Install 💾
• Usage

• Sub-commands

  • timeit 🕐
  • insert-fake-data
  • insert-json
  • insert-blob
  • run-spec
  • run-crate
  • run-track

• Development ☢

1. To quickly produce sample data. Often if someone reports an issue sample data is required to be able to reproduce it. insert-fake-data and insert-json address this problem.
2. To benchmark queries & compare runtime across Crate versions. timeit 🕐, run-spec and run-track can be used to get runtime statistics of queries. These tools focus on response latencies. Being able to benchmark throughput is NOT a goal of cr8. Similarly, being able to simulate real-world use cases is also NOT a goal of cr8.

Python >= 3.6 is required to use the command line tools.

Install them using pip:

python3.6 -m pip install --user cr8

The main binary is called cr8 which contains a couple of sub-commands.

Use cr8 -h or cr8 <subcommand> -h to get a more detailed usage description.

The included sub-commands are described in more detail below:

A tool that can be used to measure the runtime of a given SQL statement on a cluster:

>>> echo "select name from sys.cluster" | cr8 timeit --hosts localhost:4200
Runtime (in ms):
    mean:    ... ± ...
    min/max: ... → ...
Percentile:
    50:   ... ± ... (stdev)
    95:   ...
    99.9: ...

A tool that can be used to fill a table with random data. The script will generate the records using faker.

For example given the table as follows:

create table x.demo (
    id int,
    name string,
    country string
);

The following command can be used to insert 1000 records:

>>> cr8 insert-fake-data --hosts localhost:4200 --table x.demo --num-records 200
Found schema:
{
    "country": "string",
    "id": "integer",
    "name": "string"
}
Using insert statement:
insert into "x"."demo" ("country", "id", "name") values (?, ?, ?)
Will make 1 requests with a bulk size of 200
Generating fake data and executing inserts
<BLANKLINE>

It will automatically read the schema from the table and map the columns to faker providers and insert the give number of records.

(Currently only top-level columns are supported)

insert-json can be used to insert records from a JSON file:

>>> cat tests/demo.json | cr8 insert-json --table x.demo --hosts localhost:4200
Executing inserts: bulk_size=1000 concurrency=25
Runtime (in ms):
    mean:    ... ± 0.000

Or simply print the insert statement generated from a JSON string:

>>> echo '{"name": "Arthur"}' | cr8 insert-json --table mytable
('insert into mytable ("name") values (?)', ['Arthur'])
...

A tool to upload a file into a blob table:

>>> cr8 insert-blob --hosts localhost:4200 --table blobtable specs/sample.toml
http://.../_blobs/blobtable/2917773e74ff46d08f399435ed9b99afb9ed34bd

A tool to run benchmarks against a cluster and store the result in another cluster. The benchmark itself is defined in a spec file which defines setup, benchmark and teardown instructions.

The instructions itself are just SQL statements (or files containing SQL statements).

In the specs folder is an example spec file.

Usage:

>>> cr8 run-spec specs/sample.toml localhost:4200 -r localhost:4200
# Running setUp
# Running benchmark
<BLANKLINE>
## Running Query:
   Statement: select count(*) from countries
   Concurrency: 2
   Iterations: 100
Runtime (in ms):
    mean:    ... ± ...
    min/max: ... → ...
Percentile:
    50:   ... ± ... (stdev)
    95:   ...
    99.9: ...
...
## Skipping (Version ...
   Statement: ...
# Running tearDown
<BLANKLINE>

-r is optional and can be used to save the benchmark result into a cluster. A table named benchmarks will be created if it doesn't exist.

Writing spec files in python is also supported:

>>> cr8 run-spec specs/sample.py localhost:4200
# Running setUp
# Running benchmark
...

Launch a Crate instance:

> cr8 run-crate 0.55.0

This requires Java 8.

run-crate supports chaining of additional commands using --. Under the context of run-crate any host urls can be formatted using the {node.http_url} format string:

>>> cr8 run-crate latest-stable -- timeit -s "select 1" --hosts '{node.http_url}'
 # run-crate
===========
<BLANKLINE>
...
Starting Crate process
Crate launched:
    PID: ...
    Logs: ...
    Data: ...
<BLANKLINE>
...
Cluster ready to process requests
<BLANKLINE>
<BLANKLINE>
# timeit
========
<BLANKLINE>
<BLANKLINE>
<BLANKLINE>
<BLANKLINE>

In the above example timeit is a cr8 specific sub-command. But it's also possible to use arbitrary commands by prefixing them with @:

cr8 run-crate latest-nightly -- @http '{node.http_url}'

A tool to run .toml track files. A track is a matrix definition of node version, configurations and spec files.

For each version and configuration a Crate node will be launched and all specs will be executed:

>>> cr8 run-track tracks/sample.toml
# Version:  latest-testing
## Starting Crate latest-testing, configuration: default.toml
### Running spec file:  sample.toml
# Running setUp
# Running benchmark
...

To get a sandboxed environment with all dependencies installed use venv:

python -m venv .venv
source .venv/bin/activate

Install the cr8 package using pip:

python -m pip install -e .

Run cr8:

cr8 -h

Tests are run with python -m unittest