https://github.com/yeasy/ckeeper
Container application health keeper
https://github.com/yeasy/ckeeper
Last synced: 3 months ago
JSON representation
Container application health keeper
- Host: GitHub
- URL: https://github.com/yeasy/ckeeper
- Owner: yeasy
- License: apache-2.0
- Created: 2016-08-20T09:57:07.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2016-08-24T13:19:37.000Z (almost 9 years ago)
- Last Synced: 2025-01-03T12:22:35.795Z (5 months ago)
- Language: Go
- Size: 1.14 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ckeeper
[](https://travis-ci.org/yeasy/ckeeper)
[](https://goreportcard.com/report/github.com/yeasy/ckeeper)
Keep the health of container applications.
ckeeper can be deployed on each docker host, and it will automatically check the health status, e.g., validate if the web service online, and run specified operations on `unhealthy` containers, e.g., `restart`.
The default configuration is in [ckeeper.yaml](ckeeper.yaml).
## Features
* Automatically watch the health status of application as a nanny.
* Support container selection based on flexible filtering options.
* Support customized healing actions for unhealthy applications.
## Usage
There are two ways to use ckeeper: docker image (recommended) and local installation.
### Run in container
```sh
$ docker run \
-v YOUR_CONFIG_FILE:/cmonit/ckeeper.yaml \
yeasy/ckeeper \
start --logging-level=debug
```
### Local run
```sh
$ make run
```
## Configuration File
### logging
* `level`: Set output logging message level: `debug|info|warning|error|critical`.
### host
* `daemon`: The daemon url of the container host
### check
* `retries`: Each check will run how many tries before thinking it unhealthy.
* `interval`: Seconds between two check.
### Rules
Define how to decide unhealthy status and what action to take.
You can put multiple rules here, each rule will be processed in order.
For each rule:
* `option`: Used to select containers to check health, following the [ListOption](https://docs.docker.com/engine/reference/api/docker_remote_api_v1.24/#list-containers).
* `target`: Think as healthy if the target cmd's running result is true.
* `action`: What action to take if the target is not met. If the target is not set, then take action by default. Valid action values include:
* `start`: Start the container
* `restart`: Restart the container
* `stop`: Stop the container
* `pause`: Pause the container
* `unpause`: Un-pause the container
* `rm`: Remove the container
* `print`: Print the information of the matched container
* `ignore`: Take no action
The target cmd can be customized easily to achieve flexible health as the user needs.
E.g., the following rule will keep every running web containers serving normally on port 80, otherwise, restart it.
```yaml
rule_web:
option:
All: true
Filters:
status:
- running
target: '[[ `curl -sL -w "%{http_code}\\n" CONTAINER:80 -o /dev/null` == "200" ]]'
action: "restart"
```
The following rule will start all containers in `exited` or `paused` status.
```yaml
rule_start:
option: # container filter
All: true
Filters:
status:
- exited
- paused
action: "start"
```
## Testing
```sh
$ make test
```
## TODO
* Add more docs and test cases.
* Design more powerful rule engine.