The SingleStoreDB Dev Image is the fastest way to develop with SingleStore on your laptop or in a CI/CD environment (including Github Actions). This Docker image is not supported for production workloads or benchmarks so please keep that in mind when using it.
If you have any questions or issues, please file an issue on the GitHub repo or our forums.
- How to run the Docker image?
- How to open a SQL shell?
- How to access the SingleStore Studio UI?
- Where can I learn how to use SingleStoreDB?
- How to access the Data API?
- How to use WebAssembly (Wasm) backed functions?
- How to use Docker volumes for persistent storage?
- How to initialize this container with a SQL file?
- How to set SingleStoreDB Global Variables?
- How to use a specific SingleStoreDB version?
- How to use this container in a CI/CD environment?
- How to upgrade from
singlestore/cluster-in-a-box
? - Apple Silicon (M1/M2 chips) performance notes
This Docker image allows up to 4 leaf nodes to be run on a machine (host) that has up to 32 cores and up to 128GB of RAM without the need for a license.
When running a Docker container, either with or without a free edition license, on a machine with:
- More than 32 cores, include the
--cpus=32
option. - More than 128GB of RAM, include the
--memory=128g
option.
Refer to What are license units and how do they apply to my cluster? for more information on license units and how they are calculated.
Note: We recommend using an explicit image version tag whenever possible from the list of image tags and/or the changelog.
See next section for Apple Silicon (M1/M2 chips) instructions.
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR SINGLESTORE ROOT PASSWORD" \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
ghcr.io/singlestore-labs/singlestoredb-dev:latest
If you want to configure a specific license, simply pass it as an environment variable:
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR SINGLESTORE ROOT PASSWORD" \
-e SINGLESTORE_LICENSE="YOUR SINGLESTORE LICENSE" \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
ghcr.io/singlestore-labs/singlestoredb-dev:latest
First, make sure you are using Docker desktop which supports the latest virtualization technology on Apple Silicon machines.
Second, you must enable "Use Virtualization framework" and "Use Rosetta for x86_64/amd64 emulation on Apple Silicon" in Docker Desktop preferences. If you do not do this, SingleStore will run very slowly and consume a lot of power on your Mac. You can find these settings under Docker Desktop > Preferences > General.
Third, run the following command to start SingleStore. Note the --platform linux/amd64
instruction which is required when running on Apple Silicon.
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR SINGLESTORE ROOT PASSWORD" \
--platform linux/amd64 \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
ghcr.io/singlestore-labs/singlestoredb-dev:latest
Windows Powershell and Command Prompt (CMD) work best if you run the command on a single line. Alternatively you can use backticks for multi-line strings in Powershell. Either way, the following single-line version of the command will work on Windows.
docker run -d --name singlestoredb-dev -e ROOT_PASSWORD="YOUR SINGLESTORE ROOT PASSWORD" -p 3306:3306 -p 8080:8080 -p 9000:9000 ghcr.io/singlestore-labs/singlestoredb-dev:latest
The image includes a shell which you can run interactively using docker exec
like so:
docker exec -it singlestoredb-dev singlestore -p
The above command will prompt you for the root password. You can also provide the root password at the command line immediately after the -p
flag like so:
docker exec -it singlestoredb-dev singlestore -pYOUR_ROOT_PASSWORD
You can also connect to SingleStore using any MySQL compatible client on your own machine using the following connection details:
Key | Value |
---|---|
Host | 127.0.0.1 |
Port | 3306 |
Username | root |
Password | YOUR_ROOT_PASSWORD |
SingleStore Studio is a convenient way to manage SingleStoreDB and run queries via a browser based UI. Studio runs by default on port 8080 in the container. Assuming you have forwarded port 8080 to your local machine, you can access studio at http://localhost:8080.
When opening Studio you will see a login screen. Use the username root
and the ROOT_PASSWORD
you set when starting the container.
Now that you have SingleStore running, please check out the following sections of our official documentation for guides on what to do next.
In addition to supporting the MySQL Protocol, SingleStore also has a JSON over HTTP protocol called the Data API which you can access at port 9000 in the container. Assuming you have forwarded port 9000 to your local machine, the following curl command demonstrates how you can use the Data API:
~ ➜ curl -s -XPOST -H "content-type: application/json" -d '{ "sql": "select 1" }' root:YOUR_ROOT_PASSWORD@localhost:9000/api/v1/query/rows
{
"results": [
{
"rows": [
{
"1": 1
}
]
}
]
}
Note For more information on how to use the Data API please visit the documentation.
The Code Engine feature in SingleStoreDB supports creating functions (UDFs and TVFs) using code compiled to WebAssembly (Wasm). This feature supports any language that can compile to the Wasm core specification, which allows you to create UDFs in a language of your choice using existing code libraries and run them in a sandboxed environment for enhanced security. It uses a linear memory model and provides hard memory protection boundaries.
This Docker image has Wasm functions enabled by default. You can learn how to compile and load Wasm UDFs and UDAs into SingleStoreDB on our docs.
Note If you are specifying a Singlestore version at runtime, your data will always be overwritten when the container restarts each time, as the container will always attempt to re-download and re-initialise the installed version on each run.
To get around this, build a custom image with your specific version.
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR ROOT PASSWORD" \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
-v my_cool_volume:/data \
ghcr.io/singlestore-labs/singlestoredb-dev
After creating the container with a volume, you can re-create the container using the same volume to keep your data around. This can be used to upgrade SingleStore to new versions without loosing your data. Keep in mind that SingleStoreDB does not support downgrading. Make sure to take a backup of the volume before running the upgrade.
Note In order to mount a host volume to the
/data
directory, you will need to chown the volume to UID=999 and GID=998 before mounting it. The volume will be initialized automatically if empty. Host volumes are only supported by the/data
directory.
This Docker image has a number of volume mount points in addition to /data
. Here is a table outlining each of the mount points along with roughly their contents:
mount path | description |
---|---|
/data | All of the data, config, and cache for the SingleStoreDB cluster. |
/logs | All of the tracelog files containing information that can help debug the cluster or observe it's current behavior. |
/server | The installation directory containing server binaries and other installation state. |
When this docker image starts for the first time it will check to see if an init.sql
file exists in its filesystem. The default location is /init.sh
, but it can be customized via the INIT_SQL
environment variable. If init.sql
is found, the container will run it against the database as soon as SingleStoreDB is ready.
One way to do this is mounting a init.sql
from your machine into the container using the -v
flag. Here is an example of doing this:
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR ROOT PASSWORD" \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
-v ${PWD}/test/init.sql:/init.sql \
ghcr.io/singlestore-labs/singlestoredb-dev
Replace ${PWD}/test/init.sql
with an absolute path to the SQL file you want to initialize SingleStore with.
Note
/init.sql
will only be run once. If you want to run it again you will need to delete the file/data/.init.sql.done
and then restart the container.
SingleStoreDB can be configured through the use of global variables which you can find in our documentation here. These variables can be set using environment variables when running the SingleStoreDB Dev Image using the prefix SINGLESTORE_SET_GLOBAL_
.
For example, if you want to set default_partitions_per_leaf
to 1, you would do this:
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR ROOT PASSWORD" \
-e SINGLESTORE_SET_GLOBAL_DEFAULT_PARTITIONS_PER_LEAF=1 \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
ghcr.io/singlestore-labs/singlestoredb-dev
Multiple environment variables can be specified if you want to configure multiple global variables in SingleStoreDB.
If you specify a variable which is not supported by SingleStoreDB, the image will fail to start. You can see the full error message by inspecting the failed docker container's logs using docker log
.
The SingleStoreDB Dev Image uses the latest SingleStoreDB version available in the managed service by default. If you would prefer to use another SingleStoreDB version, you will need to either build a custom version of this image or specify the version at runtime by following the tutorials below.
You can use the version numbers in the first column of the following table in order to run a specific version of SingleStoreDB. If you want to use a particular patch version, just specify that version instead.
SINGLESTORE_VERSION |
description | |
---|---|---|
8.9 | SingleStoreDB Self-Managed 8.9, see changelog for latest version | changelog |
8.7 | SingleStoreDB Self-Managed 8.7, see changelog for latest version | changelog |
8.5 | SingleStoreDB Self-Managed 8.5, see changelog for latest version | changelog |
8.1 | SingleStoreDB Self-Managed 8.1, see changelog for latest version | changelog |
8.0 | SingleStoreDB Self-Managed 8.0, see changelog for latest version | changelog |
7.8 | SingleStoreDB Self-Managed 7.8, see changelog for latest version | changelog |
7.6 | SingleStoreDB Self-Managed 7.6, see changelog for latest version | changelog |
The script /scripts/switch-version.sh
can be used to easily build a custom version of this image. The fastest way to do this is using Docker build like so:
cat <<EOF | docker build -f - -t singlestoredb-dev:custom .
FROM ghcr.io/singlestore-labs/singlestoredb-dev
RUN /scripts/switch-version.sh SINGLESTORE_VERSION SINGLESTORE_LICENSE
EOF
Make sure to replace SINGLESTORE_VERSION
and SINGLESTORE_LICENSE
with the SingleStore version you want to use (see table above) as well as your license key. After running this command, you will have a new docker image called singlestoredb-dev:custom
with the specific version of SingleStoreDB installed and ready to use.
In order to use a specific version of SingleStoreDB at runtime, you can start the Docker container with the SINGLESTORE_VERSION
environment variable set.
Warning This method will result in the container taking much longer to start (roughly a minute) because it has to download and install SingleStoreDB each time. For this reason, we recommend building a custom version of this Docker image using the instructions above.
Here is an example of using the SINGLESTORE_VERSION
environment variable to run SingleStoreDB 8.9:
docker run \
-d --name singlestoredb-dev \
-e ROOT_PASSWORD="YOUR ROOT PASSWORD" \
-e SINGLESTORE_VERSION="8.9" \
-p 3306:3306 -p 8080:8080 -p 9000:9000 \
ghcr.io/singlestore-labs/singlestoredb-dev
Note You can mount
/server
into a Docker volume to preserve the installed SingleStoreDB server binaries if you are unable to use the custom image method. This will increase subsequent startup performance at the expense of complexity.
This Docker image defines a healthcheck which runs every 5 seconds. Any CI/CD system or container runtime which respects the healthcheck should automatically wait for SingleStore to be running and healthy.
Here is an example workflow which runs SingleStore as a container service and queries it from the job.
name: my-workflow
on: push
jobs:
my-job:
runs-on: ubuntu-latest
needs: build-image
services:
singlestoredb:
image: ghcr.io/singlestore-labs/singlestoredb-dev
ports:
- 3306:3306
- 8080:8080
- 9000:9000
env:
ROOT_PASSWORD: test
# Optional:
# SINGLESTORE_LICENSE: ${{ secrets.SINGLESTORE_LICENSE }}
steps:
- name: sanity check using mysql client
run: |
mysql -u root -ptest -e "SELECT 1" -h 127.0.0.1
Here is an example workflow which runs SingleStore as a service and queries it from the job. Unfortunately Gitlab does not support Docker healthchecks for services, so additional logic must be added to wait for SingleStore to be ready. There is a three year old issue to address this problem in Gitlab, so hopefully this can be simplified eventually.
Note You can add your SingleStore license key to Gitlab secrets under the key
SINGLESTORE_LICENSE
.
image: debian
stages:
- test
variables:
ROOT_PASSWORD: test
# Optional:
# SINGLESTORE_LICENSE: $SINGLESTORE_LICENSE
testing:
stage: test
services:
- name: ghcr.io/singlestore-labs/singlestoredb-dev:latest
alias: singlestoredb-dev
script:
- apt update
- apt install -y mariadb-client curl
- curl -sI localhost:8080 --retry 30 --retry-connrefused --retry-delay 1
- mysql -u root -ptest -h singlestoredb-dev -e "create database foo"
- mysql -u root -ptest -h singlestoredb-dev -e "create table foo.bar (id int)"
- mysql -u root -ptest -h singlestoredb-dev -e "insert into foo.bar values (1),(2),(3)"
- mysql -u root -ptest -h singlestoredb-dev -e "select * from foo.bar"
Before this image existed, there was another Docker Image called singlestore/cluster-in-a-box
. The docker run command for the previous image looked something like this:
docker run -i --init \
--name singlestore-ciab \
-e LICENSE_KEY=${LICENSE_KEY} \
-e ROOT_PASSWORD=${ROOT_PASSWORD} \
-p 3306:3306 -p 8080:8080 \
singlestore/cluster-in-a-box
The differences between the old image and the new image are the following:
- The image no longer needs to be initialized before you can use it
- Startup time is much better - roughly 5 seconds with the new image versus a minute with the old image
- The Data API and External Functions features are enabled by default
- Upgrade between versions is supported and tested (downgrade is not supported)
- The new image is distributed through the Github Container Repository rather than the Docker Hub
In all cases we recommend using the new image unless you need to run a older version of SingleStore which has not been released in singlestoredb-dev-image
.
In order to support running SingleStoreDB on Apple Silicon many of our performance optimizations are disabled. This can result in unexpectedly bad performance, especially during recovery (restarting SingleStoreDB) and when running queries for the first time.
To tune this performance impact (either faster or slower) you can change the number of cores and amount of RAM allocated to the Docker virtual machine by following the documentation here.