DB Connection Issues - Upgrade - 2.9.0/2.9.1

jiriteach commented 3 years ago

I've updated to 2.9.0 and since the update, the app is reporting is unhealthy with connection issues to the db. Existing rules are working but I cannot login to NPM.

App logs are showing - error connect ECONNREFUSED

Looks like way connection approach to the db has changed. I am using yobasystems/alpine-mariadb:latest. I will try and revert the db but I am trying to do this without loosing my existing rules.

Anyone else seeing similar issues? Have removed existing containers and re-pulled without success.

My Docker compose - _version: '3' services: app: image: 'jc21/nginx-proxy-manager:latest' restart: always ports:

'80:80'
'81:81'
'443:443' environment: DB_MYSQL_HOST: "db" DB_MYSQL_PORT: 3306 DB_MYSQL_USER: "DBUSERNAME" DB_MYSQL_PASSWORD: "DBPASSWORD" DB_MYSQL_NAME: "DBNAME" volumes:
/home/pi/Docker/npm/data:/data
/home/pi/Docker/npm/letsencrypt:/etc/letsencrypt db: image: 'yobasystems/alpine-mariadb:latest' restart: always environment: MYSQL_ROOT_PASSWORD: ROOTPASSWORD MYSQL_DATABASE: DBNAME MYSQL_USER: DBUSERNAME MYSQL_PASSWORD: DBPASSWORD volumes:
/home/pi/Docker/npm/data/mysql:/var/lib/mysql_

jc21 commented 3 years ago

This shouldn't be related to the 2.9.0 release, as the node packages are the same as previous releases. Though I have been wrong before.

The CI test suite also brings up a MySQL database and tests connectivity. This might be an issue with arm (I'm guessing, since you're using yobasytems?)

v.2.9.1 is out. Test with that as well?

jiriteach commented 3 years ago

Found the issue - Its an issue with 2.9.0 and 2.9.1 and alpine-mariadb 10.5.9 - https://hub.docker.com/r/yobasystems/alpine-mariadb/tags?page=1&ordering=last_updated. I reverted to 10.4.17 and no problem. Everything is working as expected again with 2.9.1 and alpine-mariadb 10.4.17. All my existing rules etc. in place.

henkisdabro commented 3 years ago

Thanks for this solution, however when I downgraded I ended up with errors on my side, such as:

InnoDB: Unsupported redo log format. The redo log was created with MariaDB 10.5.8.

That said, I only had one host in my setup, so I rebuilt the entire stack using the new "quick setup" format anyways to get rid of the issues and start fresh.

jiriteach commented 3 years ago

@henkisdabro - I didnt see this but I cleared my logs before which its probably why so it created new logs.

jiriteach commented 3 years ago

Closing as not required anymore.

NginxProxyManager / nginx-proxy-manager

DB Connection Issues - Upgrade - 2.9.0/2.9.1 #1077