Skip to content

DM Chaos

DM Chaos #8115

Workflow file for this run

name: DM Chaos
on:
schedule:
- cron: '0 17-23 * * *' # run at minute 0 every hour from 01:00 ~ 07:00 UTC+8
workflow_dispatch:
inputs:
pr:
description: 'Which PR do you want to trigger'
required: true
default: ''
# See: https://docs.github.com/en/actions/reference/workflow-syntax-for-github-actions#concurrency.
concurrency:
group: ${{ github.ref }}-${{ github.workflow }}
cancel-in-progress: true
# A workflow run is made up of one or more jobs that can run sequentially or in parallel
jobs:
# This workflow contains a single job called "base"
base:
# The type of runner that the job will run on
runs-on: ubuntu-20.04
timeout-minutes: 50
strategy:
fail-fast: false
matrix:
chaos-obj:
[
"pod-failure-dm",
"pod-kill-dm",
"network-partition-dm",
"network-emulation-dm",
"io-chaos-dm",
]
# Steps represent a sequence of tasks that will be executed as part of the job
steps:
# Set up Go for building DM
- name: Set up Go env
uses: actions/setup-go@v3
with:
go-version: '1.23'
- name: Print Go version
run: go version
# Checks-out your repository under $GITHUB_WORKSPACE, so your job can access it
- name: Check out code
uses: actions/checkout@v2
- name: Check out code by workflow dispatch
if: ${{ github.event.inputs.pr != '' }}
uses: actions/checkout@v2
with:
ref: refs/pull/${{ github.event.inputs.pr }}/head
- name: Cache go modules
uses: actions/cache@v2
with:
path: ~/go/pkg/mod
key: ${{ runner.os }}-ticdc-${{ hashFiles('go.sum') }}
- name: Cache Tools
id: cache-tools
uses: actions/cache@v2
with:
path: tools/bin
key: ${{ runner.os }}-ticdc-tools-${{ hashFiles('tools/check/go.sum') }}
- name: install k3s
run: |
curl -fsSL https://get.k3s.io | sh -s - --write-kubeconfig-mode 644 \
"${k3s_disable_command:---disable}" metrics-server \
"${k3s_disable_command:---disable}" traefik \
--flannel-backend=none \
--docker
shell: bash
- name: Export KUBECONFIG environment variable
run: |
echo 'KUBECONFIG=/etc/rancher/k3s/k3s.yaml' >> $GITHUB_ENV
shell: bash
- name: Print cluster information
run: |
kubectl config view
kubectl cluster-info
kubectl get nodes
kubectl get pods -n kube-system
kubectl get sc
kubectl version
helm version
# Disable AppArmor for MySQL, see https://github.com/moby/moby/issues/7512#issuecomment-61787845
- name: Disable AppArmor for MySQL
run: |
sudo ln -s /etc/apparmor.d/usr.sbin.mysqld /etc/apparmor.d/disable/
sudo apparmor_parser -R /etc/apparmor.d/usr.sbin.mysqld
- name: Build DM binary
run: make dm-master dm-worker dmctl dm-chaos-case
# NOTE: we also copy config files into `bin` directory,
# so we only need to send `bin` as the context into docker daemon when building image.
- name: Build DM docker image
run: |
cp -r $GITHUB_WORKSPACE/dm/chaos/cases/conf/ $GITHUB_WORKSPACE/bin/
docker build -f $GITHUB_WORKSPACE/dm/chaos/manifests/Dockerfile -t dm:chaos $GITHUB_WORKSPACE/bin
docker image list
# Set up upstream instances
- name: Set up sources
run: |
kubectl apply -f $GITHUB_WORKSPACE/dm/chaos/manifests/sources.yaml
kubectl get -f $GITHUB_WORKSPACE/dm/chaos/manifests/sources.yaml
kubectl describe -f $GITHUB_WORKSPACE/dm/chaos/manifests/sources.yaml
- name: Wait for sources ready # kubectl wait --all not working
run: |
kubectl wait --for=condition=Ready pod/mysql57-0 --timeout=300s || true
kubectl wait --for=condition=Ready pod/mysql8-0 --timeout=300s || true
kubectl wait --for=condition=Ready pod/mariadb-0 --timeout=300s || true
sleep 10
echo show pvc
kubectl get pvc -l app=sources -o wide
echo show pv
kubectl get pv -o wide
echo show svc
kubectl get svc -l app=sources -o wide
echo show sts
kubectl get sts -l app=sources -o wide
echo show po
kubectl get po -l app=sources -o wide
echo describe po
kubectl describe po -l app=sources
echo describe pvc
kubectl describe pvc -l app=sources
kubectl wait --for=condition=Ready pod/mysql57-0 --timeout=0s
kubectl wait --for=condition=Ready pod/mysql8-0 --timeout=0s
kubectl wait --for=condition=Ready pod/mariadb-0 --timeout=0s
# Set up downstream TiDB instance (deploy a TiDB with mockTiKV, not a TidbCluster managed by TiDB-operator)
- name: Set up TiDB
run: |
kubectl apply -f $GITHUB_WORKSPACE/dm/chaos/manifests/tidb.yaml
kubectl get -f $GITHUB_WORKSPACE/dm/chaos/manifests/tidb.yaml
kubectl describe -f $GITHUB_WORKSPACE/dm/chaos/manifests/tidb.yaml
- name: Wait for TiDB ready
run: |
kubectl wait --for=condition=Ready pod/tidb-0 --timeout=300s || true
echo show pvc
kubectl get pvc -l app=tidb -o wide
echo show pv
kubectl get pv -o wide
echo show svc
kubectl get svc -l app=tidb -o wide
echo show sts
kubectl get sts -l app=tidb -o wide
echo show po
kubectl get po -l app=tidb -o wide
echo describe po
kubectl describe po -l app=tidb
echo describe pvc
kubectl describe pvc -l app=tidb
kubectl wait --for=condition=Ready pod/tidb-0 --timeout=0s
- name: Set up DM-master
run: |
kubectl apply -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-master.yaml
kubectl get -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-master.yaml
kubectl describe -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-master.yaml
# NOTE: even some DM-master instances are not ready, we still continue and let chaos test cases to check again.
- name: Wait for DM-master ready
run: |
sleep 10
kubectl wait --for=condition=Ready pod -l app=dm-master --all --timeout=300s || true
echo "<<<<< show pvc >>>>>"
kubectl get pvc -l app=dm-master -o wide
echo "<<<<< show pv >>>>>"
kubectl get pv -o wide
echo "<<<<< show svc >>>>>"
kubectl get svc -l app=dm-master -o wide
echo "<<<<< show sts >>>>>"
kubectl get sts -l app=dm-master -o wide
echo "<<<<< show po >>>>>"
kubectl get po -l app=dm-master -o wide
echo "<<<<< describe po >>>>>"
kubectl describe po -l app=dm-master
echo "<<<<< describe pvc >>>>>"
kubectl describe pvc -l app=dm-master
echo "<<<<< show current log for dm-master-0 >>>>>"
kubectl logs dm-master-0 || true
echo "<<<<< show previous log for dm-master-0 >>>>>"
kubectl logs dm-master-0 -p || true
echo "<<<<< show current log for dm-master-1 >>>>>"
kubectl logs dm-master-1 || true
echo "<<<<< show previous log for dm-master-1 >>>>>"
kubectl logs dm-master-1 -p || true
echo "<<<<< show current log for dm-master-2 >>>>>"
kubectl logs dm-master-2 || true
echo "<<<<< show previous log for dm-master-2 >>>>>"
kubectl logs dm-master-2 -p || true
- name: Set up DM-worker
run: |
kubectl apply -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-worker.yaml
kubectl get -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-worker.yaml
kubectl describe -f $GITHUB_WORKSPACE/dm/chaos/manifests/dm-worker.yaml
# NOTE: even some DM-worker instances are not ready, we still continue and let chaos test cases to check again.
- name: Wait for DM-worker ready
run: |
sleep 10
kubectl wait --for=condition=Ready pod -l app=dm-worker --all --timeout=300s || true
echo "<<<<< show pvc >>>>>"
kubectl get pvc -l app=dm-worker -o wide
echo "<<<<< show pv >>>>>"
kubectl get pv -o wide
echo "<<<<< show svc >>>>>"
kubectl get svc -l app=dm-worker -o wide
echo "<<<<< show sts >>>>>"
kubectl get sts -l app=dm-worker -o wide
echo "<<<<< show po >>>>>"
kubectl get po -l app=dm-worker -o wide
echo "<<<<< describe po >>>>>"
kubectl describe po -l app=dm-worker
echo "<<<<< describe pvc >>>>>"
kubectl describe pvc -l app=dm-worker
echo "<<<<< show current log for dm-worker-0 >>>>>"
kubectl logs dm-worker-0 || true
echo "<<<<< show previous log for dm-worker-0 >>>>>"
kubectl logs dm-worker-0 -p || true
echo "<<<<< show current log for dm-worker-1 >>>>>"
kubectl logs dm-worker-1 || true
echo "<<<<< show previous log for worker-master-1 >>>>>"
kubectl logs dm-worker-1 -p || true
echo "<<<<< show current log for dm-worker-2 >>>>>"
kubectl logs dm-worker-2 || true
echo "<<<<< show previous log for dm-worker-2 >>>>>"
kubectl logs dm-worker-2 -p || true
# NOTE: we sleep a while when check members ready in cases before applying any chaos operations.
- name: Set up chaos test cases
run: |
kubectl apply -f $GITHUB_WORKSPACE/dm/chaos/manifests/cases.yaml
kubectl get -f $GITHUB_WORKSPACE/dm/chaos/manifests/cases.yaml
kubectl describe -f $GITHUB_WORKSPACE/dm/chaos/manifests/cases.yaml
sleep 60
- name: Encode chaos-mesh action
run: |
echo CFG_BASE64=$(base64 -w 0 $GITHUB_WORKSPACE/dm/chaos/manifests/${{ matrix.chaos-obj }}.yaml) >> $GITHUB_ENV
- name: Run chaos mesh action
uses: chaos-mesh/chaos-mesh-action@master
env:
CFG_BASE64: ${{ env.CFG_BASE64 }}
# check whether complete with 1m * 20 times.
- name: Wait for chaos test case complete
run: |
$GITHUB_WORKSPACE/dm/chaos/scripts/check-case.sh
- name: Setup tmate session
if: ${{ failure() }}
uses: mxschmitt/action-tmate@v3
- name: Copy logs to hack permission
if: ${{ always() }}
run: |
mkdir ./logs
kubectl get pods --no-headers -o custom-columns=":metadata.name"|grep -E "dm-"|xargs -I{} kubectl cp {}:/log/{}.log ./logs/{}.log || true
sudo chown -R runner ./logs
# Update logs as artifact seems not stable, so we set `continue-on-error: true` here.
- name: Upload logs
continue-on-error: true
uses: actions/upload-artifact@v4
if: ${{ always() }}
with:
name: chaos-base-logs.${{ matrix.chaos-obj }}
path: |
./logs
# send Slack notify if failed.
# NOTE: With the exception of `GITHUB_TOKEN`, secrets are not passed to the runner when a workflow is triggered from a forked repository.
- name: Slack notification
if: ${{ failure() }}
env:
SLACK_WEBHOOK: ${{ secrets.SLACK_NOTIFY }}
uses: Ilshidur/[email protected]
with:
args: "chaos job failed, see https://github.com/pingcap/tiflow/actions/runs/{{ GITHUB_RUN_ID }}"