Sign up to take part
Registered users can ask their own questions, contribute to discussions, and be part of the Community!
Registered users can ask their own questions, contribute to discussions, and be part of the Community!
Hi everyone,
Recently I'm having different issues related to code environment and I need some help here.
1. We always get the error:
ERROR: Could not install packages due to an EnvironmentError: [Errno 28] No space left on device
when we try to test the API on designer. Even when we have enough free disk space after we cleaned up.
2. Our devop found a huge file (25G) under code-env-cache that belongs to a dummy API test which only contains a hello world level test. However, the size of this file is unreasonable and we want to ask is there anyway to clean this code-env-cache through DSS interface (webapp) instead of do it manually in command line?
3. The test on API-designer is often unstable. We always get 2 or 3 times network error before the dev server is correctly running or getting useful error message. Why it has this behaviour?
4. I see the speed of building API is heavily depending on code environment. So what is the best practice here? Should we prepare multiple python code-env of different combination of imported packages?
Thank your help in advance .
Oscar
Hi,
For 1, you need space in /tmp too - that's a behavior of pip
2. This can only be cleand manually currently
3. Please open a support ticket, attaching an instance diagnosis, as indicated here: https://doc.dataiku.com/dss/latest/troubleshooting/obtaining-support.html
4. We indeed recommend that you rather use at most one code environment per API service rather than one per endpoint.
Hi,
For 1, you need space in /tmp too - that's a behavior of pip
2. This can only be cleand manually currently
3. Please open a support ticket, attaching an instance diagnosis, as indicated here: https://doc.dataiku.com/dss/latest/troubleshooting/obtaining-support.html
4. We indeed recommend that you rather use at most one code environment per API service rather than one per endpoint.