Open wkjung opened 2 years ago
Hi, thanks for the report!
I'd be inclined to think setup-miniconda
itself is not responsible for this. Given the context of the issue (timeouts with a lot of concurrent calls) and the error coming from libmamba
, I'd say the problem might have to do with these two things:
libmamba
not having the same retry logic as conda
Maybe there's a way to increase the timeout threshold or disable SSL verification on the libmamba side? @wolfv, have you seen an issue like this with provision-with-micromamba
?
Hmm I haven't seen this error. We could check curl documentation.
Help,please,I have the exact same error.
This is a problem for me as well. It would be great if there were some way to simply increase the timeout slightly.
Adding --insecure
and MAMBA_NO_LOW_SPEED_LIMIT=1
worked for me, but it's obviously a temporary solution
I have the same problem pls
When I used mamba to download 82 packages (315MB) concurrently from each of 8 self-hosted runners (all the runners are running in a single machine), it randomly, but with high probability, fails with the error.
Job:
Error report.
But when I stop using mamba (by removing
use-mamba
and mambaforge lines), the error does not happen. Should I release the 'timeout' limit or restrict the parallelism so that SSL connection timeout is not reached? How can I do that then?