Seems we can sometimes have some timeout error wen a lot of container are being created at the same time on a lot of different time.
We should probably profile a bit the hub and see what is the bottleneck to make it a bit more scalable.
Some idea:
enable gevent underneath flask. This is an easy fix and that could give us small boost for a low effort
Rewrite in go: Now the zflist is a think we don't really depend on any python code for the hub, It is maybe time to rewrite it in golang. This is a bigger effort then using gevent, but this is going to be more scalable for sure
Issue originally posted zero-os#34 by @zaibon
relates to https://github.com/threefoldtech/0-templates/issues/143
Seems we can sometimes have some timeout error wen a lot of container are being created at the same time on a lot of different time.
We should probably profile a bit the hub and see what is the bottleneck to make it a bit more scalable.
Some idea: