Re: API spec questions

2020-04-14 Thread Xinbin Huang
+1 on making connection IDs unique. It's confusing to have Airflow handled load balancing here. On Tue, Apr 14, 2020, 4:59 AM Kaxil Naik wrote: > +1 to make connection ids unique > > On Tue, Apr 14, 2020 at 11:59 AM Jarek Potiuk > wrote: > > > I am also supporting converting the connection to b

Re: [DISCUSS] Utility DAGs

2020-04-14 Thread Jarek Potiuk
> > Personally, I think the best solution is to create a new command line > sub-command responsible for log and/or database cleaning. Users could then > come up with their own mechanism on how to run it (trigger it when disk or > storage percentage reaches high value or simply periodically using cr

Re: API spec questions

2020-04-14 Thread Kaxil Naik
+1 to make connection ids unique On Tue, Apr 14, 2020 at 11:59 AM Jarek Potiuk wrote: > I am also supporting converting the connection to be unique. > > I've worked with similar approach long time ago (10s of years) and it was > fine then where we have not yet figured out how to scale client/ser

Re: API spec questions

2020-04-14 Thread Jarek Potiuk
I am also supporting converting the connection to be unique. I've worked with similar approach long time ago (10s of years) and it was fine then where we have not yet figured out how to scale client/server architecture and we did not have all the nice infrastructure like load/balancing, cloud serv

Re: API spec questions

2020-04-14 Thread Kamil BreguĊ‚a
Hello, This can cause big problems with idempotence. According to RFC-7231, the DELETE method should be idempotent. For example: If you want to delete items with index from 1 to 4, you should set the following request DELETE /connections/hdfs_default/4 DELETE /connections/hdfs_default/3 DELETE /c

Re: API spec questions

2020-04-14 Thread Bas Harenslak
Same here, load balancing should be responsibility of the application itself, not Airflow. +1 for making connections unique Bas On 11 Apr 2020, at 11:13, Robin Edwards mailto:r...@bidnamic.com>> wrote: On Sat, 11 Apr 2020, 01:20 QP Hou, mailto:q...@scribd.com>> wrote: It looks like we need