如何解决Dokku部署中的预接收钩拒绝错误？

Question

I'm currently in the middle of deploying an Angular4 application on a dokku instance, but, I keep on getting this error ! [remote rejected] develop -> master (pre-receive hook declined) ! [remote rejected] develop -> master (pre-receive hook declined) .

I'm deploying from a develop branch to master, so I run the deployment command as so git push mediafactory develop:master

I'm deploying via a Dockerfile, which looks like this;

FROM node:carbon

RUN mkdir -p  /opt/app
WORKDIR /opt/app

COPY package*.json ./

RUN npm install --save @angular/cli@1.3.0
RUN npm install --only=production

COPY .angular-cli.json .
COPY . .

RUN ng build --aot -prod

ENV APP_ID setYourAppId
ENV MASTER_KEY setYourMasterKey
ENV DATABASE_URI setMongoDBURI

EXPOSE 1337

CMD ["npm", "start"]

I tried running

sudo wget -O /etc/init/docker.conf https://raw.github.com/dotcloud/docker/master/contrib/init/upstart/docker.conf

sudo service docker restart

But, it doesn't seem to solve the problem.

Answer 1

I finally got the solution to the problem. I know someone may face the same problem as I had.

There are three main reasons as to why the above problem might come up.

First, If you land into this error,

remote: <app name> is currently being deployed. Exiting...
To <dokku remote>:<app name>
! [remote rejected] develop -> master (pre-receive hook declined)

You can easily solve the above problem by running these two commands from your dokku host.

$ sudo wget -O /etc/init/docker.conf https://raw.github.com/dotcloud/docker/master/contrib/init/upstart/docker.conf

$ sudo service docker restart

Secondly, as for my case, it was a problem with the permissions of the global node modules folder which would result into an infinite loop when you try to globally install the @angular/cli. Unfortunately even after providing permissions, the pre-receive hook declined error would persist. This was because of the @ngular/cli image I was using.

The error looked like this

...
/usr/local/lib/node_modules/@angular/cli/node_modules/node-sass/.node-
gyp/8.9.1
gyp WARN EACCES user "nobody" does not have permission to access the 
dev dir 
"/usr/local/lib/node_modules/@angular/cli/node_modules/node-sass/.node-
gyp/8.9.1"
gyp WARN EACCES attempting to reinstall using temporary dev dir 
"/usr/local/lib/node_modules/@angular/cli/node_modules/node-sass/.node-gyp"
...

You can solve the above error by adding the following line to your Dockerfile , if and only if the command $ npm config get prefix returns something like /usr/local , otherwise you may mess up your permissions.

RUN chown -R $(whoami) $(npm config get prefix)/lib/node_modules

Install the @angular/cli globally in your container with this command

RUN npm install -g --unsafe-perm @angular/cli

That's a work around that solves the problem.

The other problem that might cause the problem, is adding package-lock.json to your container. You might consider ignoring it too, for it may cause the same problem. The new Dockerfile hence looks like this;

FROM node:carbon

RUN mkdir -p  /opt/app
WORKDIR /opt/app

RUN chown -R $(whoami) $(npm config get prefix)/lib/node_modules

RUN npm install -g --unsafe-perm @angular/cli

COPY package.json .

RUN npm install --only=production

COPY .angular-cli.json .
COPY . .

RUN ng build --aot -prod

ENV APP_ID setYourAppId
ENV MASTER_KEY setYourMasterKey
ENV DATABASE_URI setMongoDBURI

EXPOSE 1337

CMD ["npm", "start"]

Thirdly and lastly, if you are using a Vagrant dokku installation, the pre-receive hook declined error might be caused by low memory allocation to the virtual machine.

You may edit the Vagrantfile of your dokku repo and allocate more memory to your virtual machine. For example, you may increase the memory from 1GB to 2GB by adding the following line to your Vagrantfile ;

config.vm.provider :virtualbox do |vb|
  vb.customize ["modifyvm", :id, "--memory", 2048]
end

Afterwards, you run $ vangrant reload to apply the changes you have specified in your Vagrant file.

If the problem still persists, and you are using mongoDB , you may want to checkout your mongo instances, and ensure that they are running as expected. You can check your mongo instances status by running $ dokku mongo:list on your dokku host.

Finally, If none of the above helps, and the problem still persists, you may have to destroy your dokku app and re-create a new one. TAKE EXTRA EXTRA CAUTION BEFORE DOING THIS!!!

You may checkout a tutorial I wrote about deploying 2 apps in the same docker container, with dokku