Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP

Loading…

Is this repo maintained? #185

Closed
thanpolas opened this Issue · 14 comments

8 participants

Thanasis Polychronakis AvivLander Tadeusz Wójcik Geoff Wagstaff pvencill Sascha Reuter Bulkan Evcimen Nicholas Penree
Thanasis Polychronakis

No description provided.

AvivLander

doesn't seem so

Tadeusz Wójcik

Does anyone have fork with pull requests around QoS and job shutdowns merged in and stable?

Thanasis Polychronakis

@codefather @AvivLander I am starting work on a new queueing service.

The task has been assigned by my company and we'll do it open source, so on one hand it will be finished soon enough and on the other hand feel free to chime in for designing the API.

https://github.com/verbling/kickq/blob/master/test/README.md

Tadeusz Wójcik

@thanpolas cool, thanks, will check it out for sure.

AvivLander

@thanpolas, the project looks nice! Happy to see cron jobs added. Also, I agree that functionality is more important than UI. Do you plan to have any API for job stats?

The biggest problem we have with Kue is knowing when a job has become stuck or forever delayed. Many people report issues like this with Kue, yet their exposing API is not that useful. Knowing if it has happened has become a nightmare. In addition, we wrote some methods to help clean up Redis for finished jobs and/or for jobs as described as forever delayed.

Thanasis Polychronakis

Ghost jobs is one of the problems we want to solve...

What kind of API do you have in mind for stats? REST? Node?

Geoff Wagstaff

I wrote a module called convoy which is based on a similar concept to Kue.

We use it in production at GoSquared. It's pretty basic but has methods for detecting failed / jammed jobs and removing them. Best usage examples can be found in the tests

Thanasis Polychronakis

@TheDeveloper i studied the tests and have a few questions:

  1. Why do you manually assign the job id vs having redis generate one for you?
  2. How do you pass data to a job?
Geoff Wagstaff

@thanpolas same answer to both questions - the ID is the job. For our system it's just been designed to be a queue for unique IDs, and the message that gets passed through the redis list is this unique ID.

The idea is that your consumer will take on the job, and therefore know its ID. It will then be able to look up information about that job, get the data it needs, and do the job in question. In a way, it's more reference queueing than job queuing.

Alternatively, you could encode some basic information as the ID with JSON for example, but this may have performance implications when the payload is large.

pvencill

There've been recent commits, not sure why you guys think it's abandoned.

Thanasis Polychronakis

@pvencill the repo seems abandoned considering the pending pull requests and the bugs that are crawling for too long now.

I think it's the maintainer's job to inform us of the state.

In the meantime kickq is now complete and we are in the beta testing stage. Anyone interested can check out the API as it is now stable.

Sascha Reuter

+1 @visionmedia I consider using kue as a replacement for local queues at @doctape. Would love to know if it's still maintained.

Bulkan Evcimen

@sreuter Don't know why @visionmedia provide an update on the state of this repo but @drudge is around so it is "kind of" maintained.

Me and @LOKE is trying to help out

Nicholas Penree
Owner

Kue is still being maintained.

Nicholas Penree drudge closed this
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.