juezun / deployment Goto Github PK

If I understand correctly, tmp folder is erased after reboot. So should we put that by default in other folder?

If the ownership on the folder /var/www/INGInious is not set correctly, then the application cannot save tasks data or course modifications. The deployment scirpt already does that job for you but when you copy for examples tasks from git to the folder, it would have other ownership than the required. We can document the issue or even better. Have the error shown in the app to be something related with folder ownership so that the admin will know what to to

This can be done in NGinx.

The ruby API for linting code relies on a pre-installed software called coala which at the same time relies on already installed linters In the case of pylint, coala installs it with the instalation script. However, Oclint and the java linters are not present in the dockerfile as long as I can see.

LinterService -calls-> Coala -calls-> Oclint

So we must ensure that

1. Oclint and Java Linters are installed
2. The have an environment variable named after the command that is used by coala to call

In the case of Java, it will handled here
https://github.com/JuezUN/coala-bears/blob/master/bears/java/JavaPMDBear.py#L31

In the case of Oclint, Here
https://github.com/JuezUN/coala-bears/blob/master/bears/c_languages/OclintBear.py#L9

In those files you can find the name of the excecutables so that you can include them in the container

Please take a look at

There are some warnings that we would want to check before going to prod

This could be done in a private repository where new developers can see how to setup the VPN to allow them work from home.

When you select the account trying to use the Google Auth plugin, it does not continue and times out then.

Search for a software that monitors the resources on production server. The idea of this is to have a log or some stats about the usage of them.

The use of mv command on run.sh implies that if the command fails, the files are lost forever (unless we have git and do git checkout .)

Consider using cp instead of mv to solve this issue

Cron job to Backup database and tasks every certain time.

Describe the bug / issue
When doing the deployment in a new machine, it fails as python3.5 is not correctly installed.
We should also check all deployment process to fix all other broken dependencies or something.

The deployment of Cokapi service works perfectly in a separate machine, but all C++ code does not work as python tutor says it has taken too long to run the code.

When you use the application using HTTP you will continue all requests as HTTP, so we should redirect all HTTP connections to HTTPS.

We talked about stop cloning the repos because the code is being pulled from docker hub anyways. So there is no need of doing it

Also, we need to double check the INGInious base repo on docker hub and how it is going to integrate with our code changes on every release we make

When executing the run.sh file on the cloud instance there is an error

This is the log:

make command not found

It is solved using apt install make

After refactoring grading containers, the scripts have to be updated with the new container.

Instead of using private accounts on private repositories, we should create an account and repositories to do the respective backup.

Right now, when looking into the mongo logs we see warnings. We haven't dug to much to see what is actually happening but It might really be worth taking a look at it

sshfs is a protocol for sharing files between machines using ssh. However, each time some machine is restarted a command must be ran so that the folder gets mounted. The solution for this is to make sshfs to run automatically after linux is turned on. Keep in mind that the main machine (the one with the database) has to be up BEFORE the grader machines so that when the grader machines are UP, they can mount the folder

Allow this in order to fix the issue when using Google auth plugin.

Make easier the process to update the server with the new stuff added to INGInious repo

Command to see the logs of nginx, lighttpd, webapp, docker, among others.

When you reboot the server there are some few steps to follow to make work again the instance

Please put an string on the run.sh at the end.

Something like what you told me on facebook

'Vaya a ese archivo
Y donde dice bind_Ip
Cambielo a 0.0.0.0
E inténtelo nuevamente'

You can say that information on the ending of the bash or something like that to let the user know what to do to access the db