https://github.com/bokysan/wait-for-service
Pure bash script which wait for an availability of a service
https://github.com/bokysan/wait-for-service
Last synced: 7 months ago
JSON representation
Pure bash script which wait for an availability of a service
- Host: GitHub
- URL: https://github.com/bokysan/wait-for-service
- Owner: bokysan
- License: mit
- Created: 2020-01-21T08:07:18.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2022-02-02T10:37:38.000Z (over 3 years ago)
- Last Synced: 2025-01-25T15:41:29.716Z (9 months ago)
- Language: Shell
- Size: 13 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# About
A completely architecture-independant bash script which waits for a service to be
available.
Mostly useful for Docker and Kubernetes
[init containers](https://kubernetes.io/docs/concepts/workloads/pods/init-containers/)
The script `wait-for-service`, which will wait for a given service to be up before
continuing.
It's API/parameter compatible with
[fabric8-dependency-wait-service](https://github.com/fabric8-services/fabric8-dependency-wait-service) but
since it's written in BASH and not GO:
- there's no need to include different binaries for different platforms
- more platforms are automatically supported out of the box (e.g. wherever BASH works)
## Environment variables
The following variables can be used to control the service:
| Variable | Possible values | Default value | Description |
| -------- | --------------- | ------------- | ----------- |
| `DEPENDENCY_LOG_VERBOSE` | `true`/`false` | `true` | Puts out a little extra logging into output. (like *fabric8*) |
| `DEPENDENCY_POLL_INTERVAL` | A positive integer | 2 | The interval, in seconds, between each poll of the dependency check. (like *fabric8*) |
| `DEPENDENCY_CONNECT_TIMEOUT` | A positive integer | 5 | How long to wait before the service timeouts, in seconds. (not possible with *fabric8*) |
| `SCRIPT_TIMEOUT` | A positive integer | 0 | A total time the script is allowed to run. 0 means "no timeout" |
## Usage
This utility is used as part of the [init containers](https://kubernetes.io/docs/concepts/workloads/pods/init-containers/).
Usage:
```
wait-for-service [parameters] [tcp://host:port | postgres://[user@]host[:port] | http[s]://host[:port] | ... ] [-- command args]
wait-for-service will wait for the specified service to be up and running before returning / exiting from the function.
It will optionally run the command(s) at the end of the line after the commnad completes successfully.
Available parameters:
-t | --timeout=TIMEOUT Script timeout parameter
-v | --verbose Be verbose.
Alias for DEPENDENCY_LOG_VERBOSE=true
-q | --quiet Be quiet.
Alias for DEPENDENCY_LOG_VERBOSE=false
-c | --connection-timeout=TIMEOUT Timeout before the service is deemed inaccessible (default is 5 seconds).
Alias for DEPENDENCY_CONNECT_TIMEOUT
-p | --poll-interval=INTERVAL Interval in seconds between polling retries.
Alias for DEPENDENCY_POLL_INTERVAL
-C | --colo[u]r Force colour output.
tcp://host:port Wait for the given service to be available at specified host/port. Uses Netcat.
postgres://[user@]host[:port] Wait for PostgreSQL to be available. Uses pg_isready.
http[s]://host[:port] Wait for HTTP(s) service to be ready. Uses curl.
-- COMMAND ARGS Execute command with args after the test finishes
```
An example is given below:
```yaml
spec:
initContainers:
- name: web-for-services
image: "boky/alpine-bootstrap"
command: [ 'wait-for-service', 'postgres://user@host:port', 'https://whole-url-to-service', 'tcp://host:port', ... ]
# Or, using the fabric8-compatible symlink
command: [ 'fabric8-dependency-wait-service-linux-amd64', 'postgres://user@host:port', 'https://whole-url-to-service', 'tcp://host:port', ... ]
env:
- name: DEPENDENCY_POLL_INTERVAL
value: "10"
- name: DEPENDENCY_LOG_VERBOSE
value: "false"
- name: DEPENDENCY_CONNECT_TIMEOUT
value: "10"
```
## Supported protocols
The following protocols are supported:
* **HTTP**: Anything that starts with `http://` or `https://` is considered a HTTP protocol.
[curl](https://curl.haxx.se/) is used to make a `HTTP HEAD`request. Automatically falls back
to *TCP* check if `curl` is not available.
A service is considered up if:
* It doesn't time out
* It returns a result in the range `2xx`
* **POSTGRES**: The syntax is: `postgres://[@][:]`.
The given postgres url is parsed for username, host, and port and passed to the
[pg_isready](https://www.postgresql.org/docs/10/static/app-pg-isready.html) utility.
Automatically falls back to *TCP* check if `pg_isready` is not avaiable.
The db is considered up when *pg_isready returns a zero exit code*.
* **TCP**: The syntax is: `tcp://:`. [Netcat](http://netcat.sourceforge.net/) is used to
connect to the service. Service is considered up when *netcat* successfully connects and
*return a zero exit code*. If `netcat` is not avaialbe, uses `/dev/tcp` to send open up a port.
# Similar tools
- There's [wait-for-it](https://github.com/vishnubob/wait-for-it) which seems to have a similar goal,
but a bit different syntax.
- There's [fabric8-dependency-wait-service](https://github.com/fabric8-services/fabric8-dependency-wait-service),
which was the initial inspiration for this service