Docker based 19.1 install fails to start

[J_Roeleveld]

I have followed the steps as described here:

EGroupware/egroupware

But when starting the docker containers (without the “-d” option to be able to see the output), I get the following:

egroupware-db | 2019-08-17 20:05:11 8 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:11 9 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:11 10 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:14 13 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:14 14 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:14 15 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware | /usr/bin/php7.3 -d memory_limit=-1 /usr/share/egroupware/setup/setup-cli.php --setup-cmd-database sub_command=create_db ‘domain=default’ ‘db_type=mysqli’ ‘db_host=db’ ‘db_port=3306’ ‘db_name=egroupware’ ‘db_user=egroupware’ ‘db_pass=@OyxbAbdT>OARW]l’ ‘db_root=root’ ‘db_root_pw=secret’ 'db_grant_host=“172.17.%”'
egroupware | Can not create mysqli database egroupware on db:3306 for user egroupware!
egroupware |
egroupware | Installation failed --> exiting!
egroupware |
egroupware | Retrying EGroupware installation in 3 seconds …
egroupware-nginx | 2019/08/17 20:05:16 [error] 9#9: *1 connect() failed (111: Connection refused) while connecting to upstream, client: 172.17.0.1, server: _, request: “GET /egroupware/index.php HTTP/1.1”, upstream: “fastcgi://172.17.0.4:9000”, host: "web2.antarean.org"
egroupware-db | 2019-08-17 20:05:17 18 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:17 19 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware-db | 2019-08-17 20:05:17 20 [Warning] Access denied for user ‘egroupware’@‘172.17.0.4’ (using password: YES)
egroupware | /usr/bin/php7.3 -d memory_limit=-1 /usr/share/egroupware/setup/setup-cli.php --setup-cmd-database sub_command=create_db ‘domain=default’ ‘db_type=mysqli’ ‘db_host=db’ ‘db_port=3306’ ‘db_name=egroupware’ ‘db_user=egroupware’ ‘db_pass=n6D-XN&GG9r5->aJ’ ‘db_root=root’ ‘db_root_pw=secret’ 'db_grant_host=“172.17.%”'
egroupware | Can not create mysqli database egroupware on db:3306 for user egroupware!
egroupware |
egroupware | Installation failed --> exiting!
egroupware |
egroupware | Retrying EGroupware installation in 3 seconds …

[RalfBecker]

You are attempting an update?

Where is your MySQL/MariaDB running? On the host, on an other server

What do you have as db_host in your /var/lib/egroupware/header.inc.php?

The errors you see are from not being able to connect to your database. It’s most like be caused by a user in MySQL/MariaDB is bound to an IP address or “localhost” (not 127.0.0.1, but the unix domain socket). As the later is the default for our installation, we pass the socket into the container, so you can continue to use localhost. If you used - for whatever reason - an other IP, it will probably not work, because the database sees the connection now under the docker IP. To cut that short, if your database runs on the host: use localhost for db_host in header.inc.php and docker restart egroupware.

By the way, not using -d shows the stdout, but will not detach from your terminal, so if you hit Ctrl-C, you will terminate the containers. Better use -d and and run docker-compose logs -f to tail the stderr of the containers.

Ralf

[J_Roeleveld]

For testing with the new version, I want to run a clean new installation first. There is no data present yet.
The installation uses the mariadb container created through the “docker-compose.yml” file.

I removed “-d” for debugging in a seperate terminal. I normally have multiple terminals running at the same time.

When I have a clean install running and most things working, I want to test an update.

–
Joost

[J_Roeleveld]

Further testing, when I add " - EGW_SKIP_INSTALL=true " to the environment section of “egroupware”, it skips trying to do the install itself.

When I then configure PostgreSQL as my database (newly created) (as done for all previous versions), it doesn’t try to connect at all and bounces back to the first page of setup stage.

[RalfBecker]

Hi Joost,

now I understand

I will re-test with our docker-compose.yaml file https://github.com/EGroupware/egroupware/tree/master/doc/docker

It was created before the packages and I have added a lot of stuff to it, which I added for the packages, but not constantly tested with it. I’ll try again and come back to you.

Ralf

[J_Roeleveld]

Hi Ralf,

If you could also provide a simple list of steps to take to get it setup “correctly”, eg, where to place the different files and which directories to create.
For me, most steps were trial and error. Which also could be part of the problem I am facing now.

–
Joost

[RalfBecker]

Hi Joost,

was one of those untested changes.

You need to change: DB_GRANT_HOST=172.%
So not 172.17.%, as docker-compose creates it’s own subnet and NO quotes!

Also documented that you need to create a data directory in the directory you downloaded the docker-compose.yml (or create and change the directory to something else, like our usual default /var/lib/egroupware).

Ralf

[RalfBecker]

When I then configure PostgreSQL as my database (newly created) (as done for all previous versions), it doesn’t try to connect at all and bounces back to the first page of setup stage.

Not sure what you mean with bounces back to the first page.

Where is the PostgreSQL db running, in a container created via a docker command or docker-compose or on the host itself?

EGroupware does not try to create a DB itself for PostgreSQL (never did). PostgreSQL need to allow a tcp connection to the database from 172… (docker exec -it egroupware ip -4 a). If PostgreSQL runs in a container itself it depends on how the container is created (docker-compose is by default in it’s own subnet, unless you expose it or run on the default docker network).

In general I tested yesterday with PostgreSQL 11 and was able to install a 19.1, but I had to fix an SQL error in ProjectManager, which is obviously not in the 19.1 container released on Tuesday.

Ralf

[J_Roeleveld]

Thank you, this works now.
Performance is a lot slower than what I am used to, will need to find out how to solve that.

[J_Roeleveld]

I will log a more detailed question about the install I tried for PostgreSQL.
For your info, the DB-server is a seperate host. My plan is to have a seperate machine running the docker containers and have my main webserver proxy connections to the docker container(s). I didn’t see any connection attempts to the DB-server at all.

[J_Roeleveld]

For completeness, the issue I had with PostgreSQL has been fixed.
When trying to get it working earlier, I disabled some networking options in docker itself. This caused networking problems.
I can now start testing using PostgreSQL. Will keep you updated on any issues.

[shri]

Hi can you please help,

i am exactly facing the same issues you faced earlier.

Not able to setup egroupware-docker with postgresql-11 on centos-7

docker is new for me.

currently egroupware-docker is trying to install and failing with “Can not connect to mysqli database mysql on host localhost:3306 using user root!”

what needs to be done to configure it with postgresql ?

Pl. guide.

[RalfBecker]

I’ll write something up about PostgreSQL installation in our Wiki troubleshooting section:

EGroupware/egroupware

Ralf

[hgreen]

Hi, I’m trying to install egroupware 19.1 with docker-compose. I’m facing the same problem as shri above me. To be honest if I don’t have to postgreSQL I’d prefer to not using it. I’m using the newest files from github. These are my docker-compose logs

egroupware-nginx | 2019/09/23 11:37:37 [emerg] 1#1: host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware-nginx | nginx: [emerg] host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware-nginx | 2019/09/23 11:37:42 [emerg] 1#1: host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware-nginx | nginx: [emerg] host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware-nginx | 2019/09/23 11:37:44 [emerg] 1#1: host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware-nginx | nginx: [emerg] host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8
egroupware | Fix APC(u) configuration, set apc.shm_size=128M in /etc/php/7.3/cli/conf.d/20-apcu.ini
egroupware | /usr/bin/php7.3 -d memory_limit=-1 /usr/share/egroupware/setup/setup-cli.php --setup-cmd-database sub_command=create_db ‘domain=default’ ‘db_type=mysqli’ ‘db_host=localhost’ ‘db_port=3306’ ‘db_name=egroupware’ ‘db_user=egroupware’ ‘db_pass=?A02V7WC=}-9kJ1U’ ‘db_root=root’ ‘db_root_pw=’ 'db_grant_host=localhost’
egroupware | Can not connect to mysqli database mysql on host localhost:3306 using user root! (ADOdb::Connect(localhost:3306, root, $Password, mysql) failed.)
egroupware |
egroupware | Installation failed --> exiting!
egroupware |
egroupware | Retrying EGroupware installation in 3 seconds …
egroupware | /usr/bin/php7.3 -d memory_limit=-1 /usr/share/egroupware/setup/setup-cli.php --setup-cmd-database sub_command=create_db ‘domain=default’ ‘db_type=mysqli’ ‘db_host=localhost’ ‘db_port=3306’ ‘db_name=egroupware’ ‘db_user=egroupware’ ‘db_pass=1l[FBv&iS@]emX)_’ ‘db_root=root’ ‘db_root_pw=’ 'db_grant_host=localhost’
egroupware | Can not connect to mysqli database mysql on host localhost:3306 using user root! (ADOdb::Connect(localhost:3306, root, $Password, mysql) failed.)
egroupware |

I would be grateful for some tips on where I’m going wrong. If all fails I’ll try using postgreSQL but I’d like to use the automatic install option.

Kind regards Hubert

[StefanU]

Hi Hubert.

Please ALWAYS post your own topic for your own problem. Never steal another topic (which is still closed).
You can then refer to another topic in your topic. Just copy the link into it.

Thanks and Regards
Stefan
EGroupware Community Manager

[RalfBecker]

Hi Hubert,

if I don’t have to postgreSQL I’d prefer to not using it

Using MariaDB or MySQL is the default, so that is fine.

I’m using the newest files from github.

You mean you use just our docker-compose file, and not the package installation?

egroupware-nginx | nginx: [emerg] host not found in upstream “egroupware:9000” in /etc/nginx/conf.d/default.conf:8

Somehow has Nginx been started before EGroupware, in spite of being marked as depending on egroupware container or egroupware container has been terminated.

Can not connect to mysqli database mysql on host localhost:3306 using user root! (ADOdb::Connect(localhost:3306, root, $Password, mysql) failed.)

Question is really if you use a package installation or the docker-compose file?

The error means the container is not able to talk to the database or login with the given credentials.
The docker-compose file uses a MariaDB running as container and therefore not on localhost, but host db.
Some distributions like CentOS do not start the Database by default, that could lead to your problem in a package installation.

Ralf

[hgreen]

I’ll start with saying that i’m using ubuntu 18.04. My goal is to setup colabora:code and rocket.chat to work with egroupware. I started with installing the package but all the containers didn’t start. After uncomenting in the yaml

`- EGW_DB_ROOT=root

• EGW_DB_ROOT_PW=secret`

I got the egroupware container working but rocket.chat and collabora are in different yamls so the don’t start now.
I was wondering how exactly your installation script works that the docker-compose files are in three different folders? Would it be possible to make one .yaml to have all the containers running?

Thank you so much for the quick reply

Kind Regards, Hubert