Hi Trent, One way is to deploy one standalone airflow components (scheduler, worker, broker, mysql etc.) in docker for every single user, but there needs to only one web service for user accessing, that would mean the web service should be interact with multiple other worker / scheduler for individual user, do I understand correctly ?
For Flask, how could the multi-user feature be supported ? could you help clarify ? Also, could you help share what's the reason/design behind why it doesn't support multi-user feature in airflow ? Thanks, Song ________________________________ 发件人: Trent Robbins <[email protected]> 发送时间: 2018年2月2日 3:50 收件人: [email protected] 主题: Re: About multi user support in airflow Hi Song, I would recommend using airflow in docker with individual databases for individual contributors. Does this make sense for your needs? There are other similar concepts that could also be used. I don't think there is another way to split DAGs by user permission. However, this feature doesn't seem particularly challenging to implement since it uses Flask. But who knows! Trent On Thu, Feb 1, 2018 at 19:24 Song Liu <[email protected]> wrote: > Hi, > > In production, multi-user support is needed, which means that every user > could login the airflow platform, and could manage their own dags > separately. > > But currently it seem that all the dags are managed into a single dags > folder, so for multi-user support what I need to do ? > > Many thanks for helping out. > > Thanks, > Song > -- (Sent from cellphone)
