Radek, I thought about the approach, when jobs after processing will commit offsets, but I didn't try yet (will it work?).

As of submit back I am afraid that there might be a failure that will kill container even before I understand that. I think I need to monitor the containers and their exit status codes, and in case of failure, relaunch them or say user that something went wrong.

Regards,
Uladzimir



On [email protected], Mar 28, 2017 8:05 AM wrote:

Another approach (cli-based & short lived framework) for running multiple arbitrary one-off tasks…

 

https://github.com/asamerh4/mesos-batch

 

Cheers,

Hubert

 

From: vvshvv [mailto:[email protected]]
Sent: Montag, 27. März 2017 20:53
To: [email protected]
Subject: framework for short living tasks (1-15 seconds)

 

Hi,

 

I want to run some image processing tasks on mesos cluster, but at the same time there will be a lot of tasks (1k-10k).

 

What framework do you suggest to use in such case?

 

Regards,

Uladzimir

Reply via email to