search

inensus / MPManager

Management Tool for Decentralized Utilities
GNU Affero General Public License v3.0
16 stars 20 forks source link

Small issues with local docker setup for testing #181

Closed martinjaeger closed 1 year ago

martinjaeger commented 3 years ago

Thanks a lot for open-sourcing the MPManager!

I'm trying it out with the instructions in the Readme.md, but I faced a few small challenges.

  1. When connecting to the web interface it complained about missing access rights to the following files/folders:
Website/htdocs/mpmanager/storage/logs/laravel.log
Website/htdocs/mpmanager/storage/framework/sessions/
Website/htdocs/mpmanager/storage/framework/views

A chmod a+w solved the issue, but this doesn't look like a very clean solution. Did I do something wrong?

  1. Now I've got the web interface running and I can see the login screen. Which is the default username + password? Couldn't find it in the documentation and there is also no .env file or anything similar to set it via Docker, as far as I can see.
alchalade commented 3 years ago

Hi Martin,

Issue1

Docker uses the same user to interact with the system that you fired it up with. In other words, all new directories and co. are created by that user. As far as I know, you need to

As far as I know, there are two possible solutions for it; 1 add your current user to www-data user group

  1. change the owner via chown to a user that is in the www-data group

Issue2

What you're looking for point 2, its actually in the readme see: Migrate the Database. You need to run php artisan migrate --seed inside the Laravel container that will create all database tables including an admin user with admin@admin.com and basic-password.

martinjaeger commented 3 years ago

Thanks for the quick response.

Docker uses the same user to interact with the system that you fired it up with. In other words, all new directories and co. are created by that user.

For some reason, the newly created files were owned by root even though I ran docker-compose as a normal user. Will try to look into that again.

What you're looking for point 2, its actually in the readme see: Migrate the Database. You need to run php artisan migrate --seed inside the Laravel container that will create all database tables including an admin user with admin@admin.com and basic-password.

Ok, seems like I was a bit confused by the word migrate as I thought that this would only be required if you have a previous installation with different database.

I've done that and still can't login. With above user/password combination I don't get any error message, but I don't get forwarded to the actual website. Wrong passwords result in an error message as this one:

image

Here is the log of the failed attempt with correct password:

laravel       | 172.18.0.4 - - [11/May/2021:18:24:05 +0000] "POST /api/auth/login HTTP/1.1" 200 714 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
nginx-proxy   | nginx.1    | mpmanager.local 172.18.0.1 - - [11/May/2021:18:24:05 +0000] "POST /api/auth/login HTTP/1.1" 200 473 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
laravel       | 172.18.0.4 - - [11/May/2021:18:24:05 +0000] "GET /api/settings/main HTTP/1.1" 200 533 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
nginx-proxy   | nginx.1    | mpmanager.local 172.18.0.1 - - [11/May/2021:18:24:05 +0000] "GET /api/settings/main HTTP/1.1" 200 292 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
laravel       | 172.18.0.4 - - [11/May/2021:18:24:05 +0000] "GET /api/map-settings HTTP/1.1" 200 396 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
nginx-proxy   | nginx.1    | mpmanager.local 172.18.0.1 - - [11/May/2021:18:24:05 +0000] "GET /api/map-settings HTTP/1.1" 200 155 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
laravel       | 172.18.0.4 - - [11/May/2021:18:24:05 +0000] "GET /api/ticket-settings HTTP/1.1" 200 438 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
nginx-proxy   | nginx.1    | mpmanager.local 172.18.0.1 - - [11/May/2021:18:24:05 +0000] "GET /api/ticket-settings HTTP/1.1" 200 197 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
laravel       | 172.18.0.4 - - [11/May/2021:18:24:05 +0000] "POST /api/auth/refresh HTTP/1.1" 400 470 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"
nginx-proxy   | nginx.1    | mpmanager.local 172.18.0.1 - - [11/May/2021:18:24:05 +0000] "POST /api/auth/refresh HTTP/1.1" 400 220 "http://mpmanager.local/" "Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/90.0.4430.72 Safari/537.36"

Any ideas?

alchalade commented 3 years ago

To be honest, that's the first time we're facing an issue like that.

According to the logs, the login was accepted by the back-end also Vue understood it and started to pull the base settings from the server. However, the part which refreshes/extends the JWT token on each page-load failed to re-create a new one.

Could you kindly paste both your storage/logs/laravel.log file and the response of that request refresh (in your network tab in the console) here too?

Unfortunately I was not able to reproduce that error. I installed a fresh copy(even deleted elder docker images) on Centos7 , Ubuntu20.04 and a Mac-BigSur.

I use Docker in version 20.10.02 and docker-compose in version 1.24.1.

martinjaeger commented 3 years ago

Ok, seems to be another file access rights problem. Here is the output from storage/logs/laravel.log:

[2021-05-12 04:51:09] development.ERROR: file_put_contents(/var/www/html/mpmanager/storage/framework/cache/data/a5/6f/a56f67bff77a51d486911549092b9a4f310e1df3): failed to open stream: No such file or directory {"userId":1,"exception":"[object] (ErrorException(code: 0): file_put_contents(/var/www/html/mpmanager/storage/framework/cache/data/a5/6f/a56f67bff77a51d486911549092b9a4f310e1df3): failed to open stream: No such file or directory at /var/www/html/mpmanager/vendor/laravel/framework/src/Illuminate/Filesystem/Filesystem.php:135)
[stacktrace]

Allowing write access to the cache folder did the trick. Now I can log in.

As far as I understand, docker does not necessarily use the same uids/gids as the host, so changing the group of the folders to www-data will only work if the host system is the same as the system used by the docker container (or if the gid is the same by coincidence). I'm running Manjaro, where http has gid 33. Apparently it's the same as ẁww-data in the laravel container, so we are lucky. But it would be good to make this more portable by using docker volumes instead of the local file system for any folders that need write access.

I'll probably create a PR with a suggestion later today.