[Local dev server] Workers/jobs work only sometimes, throwing `NoMethodError` for the `deliver` method when failing about cloudtasker HOT 5 CLOSED

I'll still commit a small patch to avoid these errors on v0.12. Having some guarding against nil on the local server won't hurt.

Thanks for reporting this!

from cloudtasker.

Hi @feute,

It's weird, it's as if the local server queue was filled with an empty task. Would you be able to try using v0.12.rc9 instead and let me know if the result is the same on your side?

Also, side question, do you have any logic in your application regularly clearing the cache? Or any other application running in parallel tampering with the cache?

If so, you may point Cloudtasker to a different redis database using the config.redis = { url: 'redis://localhost:6379/5' } configuration option.

from cloudtasker.

I only use redis for ActionCable. I tried adding the config line for redis to use the /5 database but it didn't do anything, I guess it's because the default is /0 while ActionCable is configured to use /1, so it doesn't interfere.

I installed v0.12.rc9 and it seemed to work better. Sometimes I get the error with the DummyWorker: I fire up the console and call perform_async on the class and I get the error most of the time, but after retrying once or twice it starts to work. I kept the config.redis line just in case.

Is it safe to use v0.12.rc9 on production? I only mean to use it as a backend for ActiveJob (at least for now).

Edit: I get the error with the job as well, but only sometimes, and it works again if I call the perform_later function again. Would it be possible that the task is being added to redis after it is being pop'd by cloudtasker? I don't know if cloudtasker ensures that the data is present on redis before calling pop, but I'd think it's something like that because it works most of the time.

from cloudtasker.

v0.12.rc9 is safe to use in production (we currently use this version at Keypup and have processed millions of jobs already).

Now for the purpose of prod you can actually use v0.11.0 as well. The issue you're facing is only related to the cloudtasker local server, which is only to be used in development. In prod it's Google Cloud Tasks that will send the tasks to your application so the issue will not arise.

Just out of curiosity, do you have only one cloudtasker process running in dev? Or do you have several applications running it? Or do you use foreman process scaling options? (foreman start web=2 worker=3 etc.)

from cloudtasker.

That's what I thought, that it could be affecting the local server only. I think I'm going to use v0.12.rc9 because it works better for me.

I do have only one cloudtasker process running. I start it with foreman start, with no extra configuration.

Anyway, if I find out the cause of this particular error (something that is on my end) I'll update this issue. For now, I'll use the 0.12.rc9 version.

Thank you!

from cloudtasker.