Skip to content

Latest commit

 

History

History
executable file
·
97 lines (67 loc) · 2.64 KB

README.md

File metadata and controls

executable file
·
97 lines (67 loc) · 2.64 KB
Raw

Watchdog

The watchdog monitors each service on the dtz cluster and notifies via Slack. Additionally, a second watchdog watches the main one.

Contents

  1. Requirements
  2. Usage

Requirements

Clone the repository and set the environment:

git clone https://github.com/iot-salzburg/dtz_watchdog.git
cd dtz_watchdog
virtualenv -p $(which python3) setup/venv
source setup/venv/bin/activate
python -m pip install -r requirements.txt

Usage

Cluster Watchdog

The Watchdog uses a slack webhook to notify about cluster issues. Therefore open in Slack a new channel, then add app, look for Incoming WebHooks, Add Configuration and select the desired Slack Channel. Then a new configuration will show the WebHook-Url in the form: https://hooks.slack.com/services/id1/id2/id3. This URL should be set as environment variable on the host. Note that the url is in quotes, so that it can be accessed better within python. Additionally set the ip-address and hostname of the cluster watchdog, and also the ip-address of the meta watchdog.

echo "SLACK_URL=https://hooks.slack.com/services/id1/id2/id3" >> .env
echo "CLUSTER_WATCHDOG_HOSTNAME=il071" >> .env
echo "SWARM_MAN_IP=192.168.48.71" >> .env
echo "META_WATCHDOG_URL=192.168.48.50" >> .env
echo "PORT=8081" >> .env
cat .env

Now, the Watchdogs can be started.

user@il071:dtz_watchdog$ python3 src/cluster-watchdog.py
user@il050:dtz_watchdog$ python3 src/meta-watchdog.py

View if the cluster is healthy http://il071:8081/.

Deployment

Adjust the settings like user name and absolute path to the repository and copy the systemd service into the proper directory:

sudo nano setup/watchdog.service
sudo cp setup/watchdog.service /etc/systemd/system/watchdog.service

An execute to enable:

sudo systemctl enable watchdog.service
sudo systemctl start watchdog
sudo systemctl status watchdog

Meta Watchdog Deployment

As there would be no notifications if the host server itself crashes, we deploy a meta watchdog, that watches only on the cluster watchdog.

Therefore, adjust the settings like user name and absolute path to the repository and copy the systemd service into the proper directory:

sudo nano setup/meta-watchdog.service
sudo cp setup/meta-watchdog.service /etc/systemd/system/meta-watchdog.service

An execute to enable:

sudo systemctl enable meta-watchdog.service
sudo systemctl start meta-watchdog
sudo systemctl status meta-watchdog

Both watchdogs should now be up and running: