I have a scenario where I want to start a server within a container and then, once it's up and running, hit it with a curl command to further configure it. Does anyone have an example of something like this?

If using windows operating system this may cause a problem.
Desite supplying a command with host and port as below:

When using command in docker-compose

    command: ["/wait-for-it.sh", "myservice:4444", "npm", "test"]

error

"Error: you need to provide a host and port to test."

Solution, download file manually by clicking 'raw' on github and downloading it.

Installing 'coreutils' and setting the alias to timeout for gtimeout does not work.

It is missing '--timeout' flag. So the code fails on line 56 or 58 where it specifies --timeout=$TIMEOUT.

I do not want to use PERL workaround as it won't be available on all machines. And the script needs to run on both Mac OS and linux alpine, so I can't simply change it to work with one or the other.

I was using wait-for-it.sh for starting service from docker-compose after RabbitMq is ready.
Though wait-for-it.sh was overriding my $HOST and $PORT variables that the service was dependent on.

I customized script a little bit to dump env variables and then return them back when executing commands after --

I suggest to add this to script or if needed I can open a PR for
as it was relatively hard to debug

In busybox commit c9720a761, a breaking change was introduced that
removes the -t flag to timeout in favor of just taking the time as
the second arg, bringing it in line with coreutils' timeout. This
commit was included in tag 1.31.1, released on 25 Oct, 2019, at
which point my integration tests that spin up a server using wait-for-it
started to fail.

i suggest wait-for-it support multiple condition
example:
./wait-for-it.sh mysql:3306 redis:6379

After updating my image to Alpine 3.7 i have noticed that wait-for-it hangs after success of finding host.

Hi,

I noticed some abnormal behavior and i need some tips if you can help me, when i run my compose file and let docker manage the networks the containers start and wait for their respectively dependencies and being available in the ports i specified but when i specify for each container network: bridge and run the compose i get this weird behavior that the container starts but never runs the init script i define it to run. in the logs it nevers shows the wait-for-it script picking up the container it was waiting for it to be live,

Any help on how to fix this

Thanks

Having quickly skimmed through the script, I couldn't find a way to wait for multiple host+port combinations to be ready. Any pointers on the best way to do this?
One way I can imagine is one wait-for-it starting another wait-for-it.

This is probably not a common situation, but in my case I need to wait for two independent databases to get ready before launching my app.

I have a Dockerfile which basis is
FROM jboss/keycloak-mysql:3.3.0.Final
Exposing the port 8080.
The wildfly comes up pretty fast, in xx seconds, the keycloak.war takes yy minutes to deploy.
My next service is relying that the keycloak war is up and running with its realms.
Using wait-for-it.sh gives clerance after xx seconds while I would want to wait for yy minutes.

The last rows in Dockerfile relying on keycloak looks lik this ( where 'sso' is my keycloak-service)

ENTRYPOINT ["/sbin/tini", "--", "/wait-for-it.sh", "sso:8080", "-t", "300", "--"]
CMD ["java", "-jar", "accounts-api-swarm.jar", "-Sinitdata" ]

So 'sso' always starts to early, the keycloak realms have not been created.
What I would like to do is to have an enhancement that works like this

ENTRYPOINT ["/sbin/tini", "--", "/wait-for-it.sh", "sso:8080/auth/realms/master/", "-t", "300", "--"]

so "sso:8080/auth/realms/master/ gives me http_status = 200 - Then and only then would like to start the java -jar accounts-api-swarm.jar Sinitdata ....

Some thoughts are here in this small repo :
https://github.com/Inkimar/wait-for

-Ingimar

I am trying to run nginx startup command in the command args but failing

bash-4.3# /tmp/wait-for-fpm.sh webapp:9000 -t 100 -- nginx -g 'daemon off;'
wait-for-fpm.sh: waiting 100 seconds for webapp:9000
wait-for-fpm.sh: webapp:9000 is available after 0 seconds
nginx -g daemon off;
nginx: invalid option: "off;"
bash-4.3# 

Can't find how the signgle quotes getting removed
The command should be as it is like nginx -g 'daemon off;' not nginx -g daemon off;

I'm having problems with running wait-for-it.sh with CouchBase Sync Gateway. Without wait-for-it.sh sync-gateway starts, with it - doesn't start.
The full description is on CouchBase forums.

Eventhough the docker cassandra container is running. I am manually testing the script on running container. However, its still printing db is up. For host I have also tried name of the container:port. I am getting same error results for that input as well

Thank you

/wait-for-it.sh: 75: /wait-for-it.sh: Syntax error: "(" unexpected (expecting ";;")
error when trying to build with a openjdk:8-jre image

OSX bash shell doesn't have a timeout command, but often (at least dev OSX machines) will have perl. Can do:

function timeout() { perl -e 'alarm shift; exec @ARGV' "$@"; }

wait-for should throw an error message if dependencies are missing.

For example the Centos7 image neither has netcat nor tcp installed. A friendly message that this packages are neccessary.

Alternatively a hint in the README should point the admin to the required packages.

When running this script in Alpine, this error occurs:

timeout: can't execute '15': No such file or directory

BusyBox v1.24.1 (2015-12-16 08:00:02 GMT) multi-call binary.

Usage: timeout [-t SECS] [-s SIG] PROG ARGS

Hi guy, when I run this script in OSX, everything will be ok. But when I deploy docker on cloud with CentOS 7, this error is shown.

timeout: can't execute '15': No such file or directory

Please help me this problem.

ERROR: for api Cannot start service api: oci runtime error: container_linux.go:247: starting container process caused "exec: "./wait-for-postgres.sh": permission denied"
ERROR: Encountered errors while bringing up the project.

on centos5, timeout command not found.
how to fix it?

echo sends a newline character which causes some server software to report an error message about an invalid HTTP request. By adding the '-n' flag the echo command can be used to open the port without sending anything.

I've created a pull request to fix this: #86

Sometimes it is needed to not just reach the endpoint, but verify that something is responding normally there. My specific use case was that I had an nginx server on which I was testing the endpoints "successfully", but they were returning 502 Bad Gateway errors because the server that it was proxying was not yet up.

Can add an option to use curl or wget to do this.

Trying all the examples in iTerm2 on my mac, I get the following error:

$ wait-for-it.sh www.google.com:80 -- echo "google is up"
wait-for-it.sh: line 53: timeout: command not found
wait-for-it.sh: timeout occurred after waiting 15 seconds for www.google.com:80
google is up

I've tried with variations such as using bash and sh and different working directories. Is this just not compatible with mac osx, El Captain?

The version where it doesn't work: GNU bash, version 4.4.23(1)-release (x86_64-apple-darwin17.5.0). After each command it shows:

readlink: illegal option -- f

The version where it works: GNU bash, version 4.2.46(2)-release (x86_64-redhat-linux-gnu)

That would be awesome !

It would be nice if --timeout=-1 would wait forever until the port is available.

The script contaminates the environment, clobbering the PORT env-var, among others. These are not-uncommon names to use when setting up services in a container.

It might be wise for the script to namespace its env-vars, like WAITFORIT_PORT

In order to use it with Docker and docker-compose,
place it alongside Dockerfile, build it into the image with exec permissions and reference it in docker-compose.yml.

- Dockerfile
- wait-for-it.sh

#Dockerfile
...
COPY wait-for-it.sh /wait-for-it.sh
RUN chmod +x /wait-for-it.sh
...

#docker-compose.yml
...
  my-service:
    build:
      context: "."
      container_name: my_container
      command: ["/wait-for-it.sh", "mysql:3306", "--", [entrypoint function]]
...

Since this script is a bash script and running it under POSIX sh will fail, it is technically incorrect for it to be named .sh.

It would be ideal if multiple services (i.e. multiple host:port combinations) could be declared such that the command is executed only after all dependent services are available.

I have a situation where the port is open but the application is still initializing. Can we add an option to wait for a HTTP resource to return?

When a preffix is used before the domain or IP, like http://localhost:8080, the script recognizes the preffix as the domain so it tries to ping without the port, like:

wait-for-it.sh: waiting 10 seconds for http://localhost
wait-for-it.sh: timeout occurred after waiting 10 seconds for http://localhost

Debian tracker reports three issues. The one with higher importance is:

The package is severely out of date with respect to the Debian Policy. The package should be updated to follow the last version of Debian Policy (Standards-Version 4.5.0 instead of 3.9.7).

Apparently #3 could be closed.

Is it possible to make this shell-independent?

/wait-for-it.sh: line 70: syntax error: unexpected "(" (expecting ";;")

This would be particularly useful to run it with alpine, which uses busybox.

Hi there! I wrote wait-for-it in order to help me orchestrate containers I operate at my day job. I thought it was a neat little script, so I published it. I assumed I would be its only user, but that's not what happened! wait-for-it has received more stars then all of my other public repositories put together. I had no idea this tool would solicit such an audience, and I was equally unprepared to carve out the time required to address my user's issues and patches. I would like to solicit a volunteer from the community who would be willing to be a co-maintainer of this repository. If this is something you might be interested in, please email me at [email protected]. Thanks!

I am using a reverse proxy where a lot of services are located under the same domain on different paths.

my-domain/api/bacon
my-domain/api/potato

And I would like to wait for each of these services individually.

Would it be possible to do something like:

wait-for-it my-domain:80/api/bacon

I've tryed some variations of it and all of them fail.

line 75: syntax error: unexpected "(" (expecting ";;")

P.S: Script running via /bin/sh

Would you mind explicitly adding a license to this script? I realize this sounds like a ridiculous request, but lawyers are annoying. ;)

Enhancement request:

• Is there any chance to have UDP test added for this script?

In my use case:

  web:
    build: .
    command: ./wait-for-it.sh mongo:27017 -t 0 -- nodemon server.js
    restart: always
    environment:
      - PORT=8096
    volumes:
      - .:/code
    ports:
      - "8096:8096"
    links:
      - mongo

  mongo:
    image: 'mongo:3.2.1'
    ports:
      - '27017:27017'
    volumes:
      - './data/db:/data/db'

I have environment variable named PORT which will be overridden by mongo's port when running wait-for-it bash.
My solution is to change PORT variable in bash to _PORT to make it looks more private.

Hello @douglas-gibbons!

Could you enable Travis?
https://travis-ci.org/github/vishnubob/wait-for-it
or
https://travis-ci.com/github/vishnubob/wait-for-it (modern)

It seems strange that all services are started, but cannot connect to the expose port...
But everything is ok via docker-compose, except docker stack deploy.

Hi,
I'm facing some issue when this script is probing a port to know if the service is up and running but sometimes the port is opened but the application is not yet ready.

Is it possible to add some sleep just after a successful check and running the command ?

https://raw.githubusercontent.com/vishnubob/wait-for-it/master/wait-for-it.sh leads to a 500: Internal Server Error. This script is so useful when using in Docker containers and pipelines, would love for it to be available again instead of having to use Git to clone repo every time.

It would be great if we could just install this from something like apt-get.

I would like to do the following steps:

• sudo apt-get install wait-for-it
• wait-for-it www.google.com:80 -- echo "google is up"

we use wait-for-it.sh heavily in CI for docker integration testing in golang. Recently a job broke due to wait-for-it.sh error. I believe the issue may have to do with something changing in timeout from alpine3.9 (works) to alpine 3.10 (not working). In both cases we were installing bash with apk.

Repro (note in vi I am copying wait-for-it from current master)

$ docker run -it --rm golang:1.12-alpine
/go # vi wait-for-it.sh
/go # chmod +x wait-for-it.sh 
/go # apk update && apk add --no-cache bash
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/community/x86_64/APKINDEX.tar.gz
v3.10.1-11-g89d0862481 [http://dl-cdn.alpinelinux.org/alpine/v3.10/main]
v3.10.1-12-ga885fe876c [http://dl-cdn.alpinelinux.org/alpine/v3.10/community]
OK: 10327 distinct packages available
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/community/x86_64/APKINDEX.tar.gz
(1/5) Installing ncurses-terminfo-base (6.1_p20190518-r0)
(2/5) Installing ncurses-terminfo (6.1_p20190518-r0)
(3/5) Installing ncurses-libs (6.1_p20190518-r0)
(4/5) Installing readline (8.0.0-r0)
(5/5) Installing bash (5.0.0-r0)
Executing bash-5.0.0-r0.post-install
Executing busybox-1.30.1-r2.trigger
OK: 15 MiB in 20 packages
/go # ./wait-for-it.sh docker:5432
timeout: unrecognized option: t
BusyBox v1.30.1 (2019-06-12 17:51:55 UTC) multi-call binary.

Usage: timeout [-s SIG] SECS PROG ARGS

Runs PROG. Sends SIG to it if it is not gone in SECS seconds.
Default SIG: TERM.
wait-for-it.sh: timeout occurred after waiting 15 seconds for docker:5432

Workaround: I was able to work around this issue by installing coreutils as well which updates timeout (continuing from example above)

/go # apk update && apk add --no-cache coreutils
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/community/x86_64/APKINDEX.tar.gz
v3.10.1-11-g89d0862481 [http://dl-cdn.alpinelinux.org/alpine/v3.10/main]
v3.10.1-12-ga885fe876c [http://dl-cdn.alpinelinux.org/alpine/v3.10/community]
OK: 10327 distinct packages available
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/main/x86_64/APKINDEX.tar.gz
fetch http://dl-cdn.alpinelinux.org/alpine/v3.10/community/x86_64/APKINDEX.tar.gz
(1/3) Installing libacl (2.2.52-r6)
(2/3) Installing libattr (2.4.48-r0)
(3/3) Installing coreutils (8.31-r0)
Executing busybox-1.30.1-r2.trigger
OK: 16 MiB in 23 packages
/go # ./wait-for-it.sh -t 15 docker:5432
wait-for-it2.sh: waiting 15 seconds for docker:5432
wait-for-it2.sh: timeout occurred after waiting 15 seconds for docker:5432

Hello!

I am going to upload wait-for-it to Debian and It would help me keep up with development if the repository was tagged. Are you able to do that please?

I was using wait-for-it on a debian docker container, and all things were great (thanks for the script!).

However, we switched to an Alpine Linux based docker image (which uses busybox) and the script stopped waiting correctly.

As it turns out timeout requires an argument flag for the time. Here's the manual from https://busybox.net/downloads/BusyBox.html

timeout
timeout [-t SECS] [-s SIG] PROG [ARGS]
Runs PROG. Sends SIG to it if it is not gone in SECS seconds. Defaults: SECS: 10, SIG: TERM.

This differs from typical linux usage:

timeout [OPTION] DURATION COMMAND [ARG]...

Changing lines 52 and 54 where timeout is used by prepending -t before the $TIMEOUT causes it to work.

Here's the working result:

wait_for_wrapper()
{
    # In order to support SIGINT during timeout: http://unix.stackexchange.com/a/57692
    if [[ $QUIET -eq 1 ]]; then
        timeout -t $TIMEOUT $0 --quiet --child --host=$HOST --port=$PORT --timeout=$TIMEOUT &
    else
        timeout -t $TIMEOUT $0 --child --host=$HOST --port=$PORT --timeout=$TIMEOUT &
    fi
    PID=$!
    trap "kill -INT -$PID" INT
    wait $PID
    RESULT=$?
    if [[ $RESULT -ne 0 ]]; then
        echoerr "$cmdname: timeout occurred after waiting $TIMEOUT seconds for $HOST:$PORT"
    fi
    return $RESULT
}

I did a quick search for a simple method to determine if busybox is being used, not sure what the best solution would be. Worst case - this post may help others in a similar situation.