tree: 5f0e456edda9f850f92ac680ceea13908baa3429 [path history] [tgz]
  1. resource/
  2. cluster.py
  3. command.py
  4. database.py
  5. Dockerfile
  6. doris-compose.py
  7. format-code.sh
  8. Readme.md
  9. requirements.txt
  10. utils.py
docker/runtime/doris-compose/Readme.md

Doris compose

Use doris compose to create doris docker compose clusters.

Requirements

1. Make sure you have docker permissions

run:

docker run hello-world

if have problem with permission denied, then add-docker-permission.

Make sure BuildKit configured in the machine. if not follow docker-with-BuildKit.

2. The doris image should contains

/opt/apache-doris/{fe, be, ms}

If don't create cloud cluster, the image no need to contains the ms pkg.

If build doris use sh build.sh --fe --be --cloud without do any change on their origin conf or shells, then its output/ satisfy with all above, then run command in doris root directory will generate such a image. If you want to pack a product that is not the output/ directory, you can modify Dockerfile by yourself.

docker build -f docker/runtime/doris-compose/Dockerfile -t <image> .

The Dockerfile default use JDK 17, for doris 2.1, 3.0, master, they all default use JDK 17.

But doris 2.0 still use JDK 8, for build 2.0 image, user need specific use JDK 8 with arg JDK_IMAGE=openjdk:8u342-jdk. Here is build 2.0 image command:

docker build -f docker/runtime/doris-compose/Dockerfile \
     --build-arg JDK_IMAGE=openjdk:8u342-jdk            \
    -t <image> .

The <image> is the name you want the docker image to have.

User can also download a doris release package from Doris Home or Doris Github, extract it, then build its image with arg OUTPUT_PATH

for example:

cd ~/tmp
wget https://apache-doris-releases.oss-accelerate.aliyuncs.com/apache-doris-3.0.5-bin-x64.tar.gz
tar xvf apache-doris-3.0.5-bin-x64.tar.gz  # after extract, there will be a directory ./apache-doris-3.0.5-bin-x64/{fe, be, ms}

# -f: the Dockerfile file
# -t: the builded image
# . : current directory, here it's ~/tmp, then output path is ~/tmp/apache-doris-3.0.5-bin-x64
docker build \
    --build-arg OUTPUT_PATH=./apache-doris-3.0.5-bin-x64 \
    -f ~/workspace/doris/docker/runtime/doris-compose/Dockerfile \
    -t my-doris:v3.0.5 \
    .

3. Install the dependent python library in ‘docker/runtime/doris-compose/requirements.txt’

PyYAML of certain version not always fit other libraries' requirements. So we suggest to use a individual environment using venv or conda.

python -m pip install --user -r docker/runtime/doris-compose/requirements.txt

if it failed, change content of requirements.txt to:

pyyaml==5.3.1
docker==6.1.3
......

Usage

Notice

Each cluster will have a directory in ‘/tmp/doris/{cluster-name}’, user can set env LOCAL_DORIS_PATH to change its directory.

For example, if user export LOCAL_DORIS_PATH=/mydoris, then the cluster's directory is ‘/mydoris/{cluster-name}’.

And cluster‘s directory will contains all its containers’s logs and data, like fe-1, fe-2, be-1, ..., etc.

If there are multiple users run doris-compose on the same machine, suggest don't change LOCAL_DORIS_PATH or they should export the same LOCAL_DORIS_PATH.

Because when create a new cluster, doris-compose will search the local doris path, and choose a docker network which is different with this path's clusters.

So if multiple users use different LOCAL_DORIS_PATH, their clusters may have docker network conflict!!!

Create a cluster or recreate its containers

python docker/runtime/doris-compose/doris-compose.py up  <cluster-name>   <image?> 
    --add-fe-num  <add-fe-num>  --add-be-num <add-be-num>
    [--fe-id <fd-id> --be-id <be-id>]
    ...
    [ --cloud ]

if it's a new cluster, must specific the image.

add fe/be nodes with the specific image, or update existing nodes with --fe-id, --be-id

For create a cloud cluster, steps are as below:

  1. Write cloud s3 store config file, its default path is ‘/tmp/doris/cloud.ini’. It's defined in environment variable DORIS_CLOUD_CFG_FILE, user can change this env var to change its path. A Example file is locate in ‘docker/runtime/doris-compose/resource/cloud.ini.example’.

  2. Use doris compose up command with option --cloud to create a new cloud cluster.

The simplest way to create a cloud cluster:

python docker/runtime/doris-compose/doris-compose.py up  <cluster-name>  <image>  --cloud

It will create 1 fdb, 1 meta service server, 1 recycler, 3 fe and 3 be.

Remove node from the cluster

python docker/runtime/doris-compose/doris-compose.py down  <cluster-name> --fe-id <fe-id>  --be-id<be-id> [--clean]  [--drop-force]

Down the containers and remove it from the DB.

For BE, if specific drop force, it will send dropp sql to FE, otherwise it will send decommission sql to FE.

If specific --clean, it will delete its data too.

Start, stop, restart specific nodes

python docker/runtime/doris-compose/doris-compose.py start  <cluster-name>  --fe-id  <multiple fe ids>  --be-id <multiple be ids>
python docker/runtime/doris-compose/doris-compose.py restart  <cluster-name>  --fe-id  <multiple fe ids>  --be-id <multiple be ids>

List doris cluster

python docker/runtime/doris-compose/doris-compose.py ls <multiple cluster names>

if specific cluster names, it will list all the cluster's nodes.

Otherwise it will just list summary of each clusters.

There are more options about doris-compose. Just try

python docker/runtime/doris-compose/doris-compose.py <command> -h 

Generate regression custom conf file

python docker/runtime/doris-compose/doris-compose.py config <cluster-name>  <doris-root-path>  [-q]  [--connect-follow-fe]

Generate regression-conf-custom.groovy to connect to the specific docker cluster.

Setup cloud multi clusters test env

steps:

  1. Create a new cluster: python docker/runtime/doris-compose/doris-compose.py up my-cluster my-image --add-fe-num 2 --add-be-num 4 --cloud
  2. Generate regression-conf-custom.groovy: python docker/runtime/doris-compose/doris-compose.py config my-cluster <doris-root-path> --connect-follow-fe
  3. Run regression test: bash run-regression-test.sh --run -times 1 -parallel 1 -suiteParallel 1 -d cloud/multi_cluster

Problem investigation

Log

Each cluster has logs in Docker in ‘/tmp/doris/{cluster-name}/{node-xxx}/log/’. For each node, doris compose will also print log in ‘/tmp/doris/{cluster-name}/{node-xxx}/log/health.out’

Up cluster using non-detach mode

python docker/runtime/doris-compose/doris-compose.py up ...   -no-detach

Developer

Before submitting code, pls format code.

bash format-code.sh