Self hosting Dokku and nextjs doesnt work in production

[gyto]

I have setup nextjs and quirrel as two instances of the dokku, each of them has its own domain and is served on:

• www.some.com
• queue.some.com

Deployment process is working fine, it register the quirrel ci and sends a job registration

nextjs APi route works fine and can retrieve route without any issues.
Locally everything works like a charm, no issues and I can easily see logs it fires the route I need. However, only when I am deploying does everything stops working for no reason. I am able to invoke the CRON job and I see in quirrel logs that it fires fine. However, it returned a status of 204 (but I am not returning any content in it)

When I invoke cron job I get these logs:

2022-08-03T14:43:02.600902477Z app[web.1]: {"level":30,"time":1659537782600,"pid":7,"hostname":"3ac4d13a0bf1","reqId":"req-5a","req":{"method":"GET","url":"/activity","hostname":"queue.some.com","remoteAddress":"172.17.0.1","remotePort":50470},"msg":"incoming request"}
2022-08-03T14:43:23.324212618Z app[web.1]: {"level":30,"time":1659537803323,"pid":7,"hostname":"3ac4d13a0bf1","reqId":"req-5b","req":{"method":"POST","url":"/queues/https%3A%2F%2Fwww.some.com%2Fapi%2Fqueues%2Fsync_ebay/%40cron","hostname":"queue.some.com","remoteAddress":"172.17.0.1","remotePort":50474},"msg":"incoming request"}
2022-08-03T14:43:23.328877486Z app[web.1]: {"level":30,"time":1659537803328,"pid":7,"hostname":"3ac4d13a0bf1","reqId":"req-5b","res":{"statusCode":204},"responseTime":3.1012649536132812,"msg":"request completed"}
2022-08-03T14:44:01.574774415Z app[web.1]: {"level":10,"time":1659537841572,"pid":7,"hostname":"3ac4d13a0bf1","module":"owl","msg":"Stale-Checker: Starting."}
2022-08-03T14:44:01.574828583Z app[web.1]: {"level":10,"time":1659537841573,"pid":7,"hostname":"3ac4d13a0bf1","module":"owl","msg":"Stale-Checker: No stale jobs found."}
2022-08-03T14:44:23.742714065Z app[web.1]: {"level":30,"time":1659537863742,"pid":7,"hostname":"3ac4d13a0bf1","reqId":"req-5c","req":{"method":"GET","url":"/activity","hostname":"queue.ocwatchguy.com","remoteAddress":"172.17.0.1","remotePort":50642},"msg":"incoming request"}

[email protected]
[email protected]
[email protected]

[Skn0tt]

The 204 response code isn't related to your Queue code, but it's the quirrel-internal response code for invoking a job. So nothing interesting there.

I see your cron schedule is 15 * * * *. Could you check the Quirrel logs during the timeframe it should trigger, and post what you see?

[Skn0tt]

oh, fair point! didn't see that. How did you deploy your Quirrel server?

[Skn0tt]

I'm thinking this might happen because only the scheduler part of quirrel is started, but not the worker part. Are you using the official Docker image?

[gyto]

I am using dokku env to deploy my main project (it's similar to the Heroku setup). It contains 2 projects both have setup domains and both accessible from the internet

domains

=====> some-quirrel domains information
       Domains app enabled:           true
       Domains app vhosts:            queue.some.com
       Domains global enabled:        true
       Domains global vhosts:         some.com
=====> some-web-app domains information
       Domains app enabled:           true
       Domains app vhosts:            www.some.com
       Domains global enabled:        true
       Domains global vhosts:         some.com

proxy

=====> some-quirrel proxy information
       Proxy enabled:                 true
       Proxy port map:                http:80:9181 http:9181:9181 https:443:9181
       Proxy type:                    nginx
=====> some-web-app proxy information
       Proxy enabled:                 true
       Proxy port map:                http:80:5000 https:443:5000
       Proxy type:                    nginx

Installation is actually pretty forward

// installed Redis and obtain Redis URL for the app
$ dokku plugin:install https://github.com/dokku/dokku-redis.git my-redis

// then installed image
$ dokku git:from-image some-quirrel ghcr.io/quirrel-dev/quirrel:sha-421e24d

// then added variables to project as is
$ dokku config:set some-quirrel PASSPHRASES=abc REDIS_URL=abc

then I obtained a token and added it to my main app as:

QUIRREL_API_URL=https://queue.some.com
QUIRREL_BASE_URL=www.some.com
QUIRREL_ENCRYPTION_SECRET=abc
QUIRREL_TOKEN=abc

[Skn0tt]

I see. I'm not too familiar with Dokku, and going into a long weekend now. I'll be able to take a closer look next week. My hunch is that the worker part of the service isn't running properly, maybe Dokku doesn't have support for that. As a workaround, you could look at https://github.com/quirrel-dev/quirrel-on-railway or a plain Docker deployment.

[gyto]

@Skn0tt you are genius, man. you're right the service did not have worker

so the solution is

dokku ps:scale my_app web=1 worker=1

So many hours of debugging and finally its working :)

[gyto]

If anyone having an issue with Dokku and Quirrel setup I would recommend checking if the server contains a worker as it was an issue for me in beginning, the deployment is actually easy and doesn't have any issues, just enable the worker :)

dokku ps:scale my_app web=1 worker=1

also big thanks to @Skn0tt for helping with the issue