app uses all cpu and can't start after upgrading to 2.16.0

[teidesu]

Describe the bug

after updating to 2.16.0 (from 2.15.0) the container won't start without any additional logging (even with -e ENVIRONMENT=dev)

System information
Please provide information on:

• Wakapi version 2.16.0
• Operating system
  • NixOS
  • Docker image ghcr.io/muety/wakapi:2.16.0
• Database Postgres

[muety]

By "can't start" you probably mean the web server doesn't come up so you can't access it?

Can you please show Wakapi's container / application logs (docker logs <container name>)? It might be telling you what it's being busy working on, which would be helpful for debugging. Also, if possible, can you also show the list of active tasks in Postgres by running the following query?

select * from pg_stat_activity;

[hypervtechnics]

I am not sure whether the poster of this gave all the information. for me the above behaviour happens when I configure an OIDC entity using environment variables from the README.

EDIT 1: After the available after x seconds line basically nothing happens.

EDIT 2: I do not use a config file and just set the four variables in addition to the ones which have been there before the update.

[teidesu]

hm. yeah, i did in fact configure an oidc provider. didn't occur to me that it might be linked 😅

yeah, same, i used all the same env vars + the oidc ones and it just. doesn't start. removing all 4 fixes the issue.

---

not only the web server doesn't start, it doesn't seem to even get to the migrations step. as stated in the op, there are no logs at all from the wakapi itself, only from the wait-for wrapper bash script.

the select * from pg_stat_activity; query returns nothing related to wakapi.

[hypervtechnics]

I have a suspicion that this is related to the env tags on the config struct for oidc. The call

wakapi/config/config.go

Line 530 in b9487c1

if err := configor.New(&configor.Config{}).Load(config, configFlag); err != nil {

is quite early in application startup and might silently fail or loop endlessly.

[anwinged]

Hi. I also encountered this behavior. I tried to configure the Authelia OIDC provider. At startup, Wakapi loads the CPU and disk completely.

• Wakapi version 2.16.0
• Operating system
  • Ubuntu 20.04.6 LTS
  • Docker image ghcr.io/muety/wakapi:2.16.0
• Database: sqlite.

I set up the OIDC using environment variables.

WAKAPI_OIDC_PROVIDER_NAME: "authelia"
WAKAPI_OIDC_PROVIDER_CLIENT_ID: "{{ wakapi_oidc_client_id }}"
WAKAPI_OIDC_PROVIDER_CLIENT_SECRET: "{{ wakapi_oidc_client_secret }}"
WAKAPI_OIDC_PROVIDER_ENDPOINT: "https://auth.vakhrushev.me/.well-known/openid-configuration"

[timoschwarzer]

I was using 2.16.0 before and just added OIDC configuration via environment variables. After that Wakapi uses all CPU available and fills up RAM until the host goes out of memory.

[muety]

Thanks for providing details! Looks like an infinite loop / recursive function calls during config parsing. Will fix this later today, sorry! ✌️

[muety]

Pushed a fix! Could somebody please double-check if it's working properly before I create a new release version? Please check the README for the new variable naming format.