Tidb-operator

tidb-operator manage multiple tidb cluster atop Kubernetes, support for the actual needs of users to apply for different specifications of resources, supports online scale up or dowm, rolling upgrades, multiple disk mount points, full / incremental migrate data to tidb cluster, all operations web.

Note: Currently only support kubernetes version is 1.6, all port ranges [10000-15000). Only provide restful api, easy to integrate into the existing web frame.

Build images

Build all docker images and push to your private docker registry, default version is latest:

make build && make push

Preparedness

Install kubernetes if have already installed, skip this step

Note: Due to GFW reasons, some installation packages and images can not be obtained, you need to download to the local upload to the specified server and then install. See: kubernetes ./kubernetes/deploy directory.

Access kubernetes dashboard: http://<NodeIP>:10281

Download

Git clone the project to $GOPATH/src/github/ffan dir

Deploy prometheus/grafana on kubernetes

make install-grafana # run this shell on kubernetes master

Access grafana: http://<NodeIP>:12802, user&password is admin/admin.

Deploy tidb-gc on kubernetes

Tidb-gc is used to recycle deleted tikvs and tidb cluster metrics in prometheus, default hostpath is /mnt, which will gc dir(/mnt/data*/) if tidb-operator starts the argument -host-path is /mnt, -mount is data.

make install-tidb-gc # run this shell on kubernetes master

Startup tidb-operator

To start on kubernetes cluster

make install-tidb-operator # run this shell on kubernetes master

Development

cd ./operator && ln -s swagger ../swagger # ln swagger to `./tidb-operator`

# beego, set the kubenetes API server endpoint before running,
# for example `export K8S_ADDRESS=http://10.213.131.54:10218` or midify restart.sh
bee run -downdoc=true
# or
./restart.sh

Access endpoint: http://127.0.0.1:12808/swagger