Closed strophy closed 1 year ago
I actually use the action again at a different point in the workflow and it works. After forcing use of the correct endpoint with endpoint: s3.eu-west-1.amazonaws.com
I noticed the difference between the working and non-working run is in the presigned request, the one that fails only has a partial X-Amz-Credential
and the X-Amz-Security-Token
is missing entirely, as well as the URL being invalid.
I noticed I use aws-actions/configure-aws-credentials@v4
before each cache step. Moving this out of the called composite action and into the calling action resulted in the cache action working correctly again, it was something to do with running the configure-aws-credentials
action twice.
This action recently stopped working for me in some cases with the following errors on restore/save. @everpcpc Is this related to OpenDAL, or am I configuring something incorrectly?
s3..amazonaws.com
makes me think it is parsing the region incorrectly?My action looks like this:
Restore (multiple hits should exist in this bucket):
Save:
Removing
endpoint: s3.amazonaws.com
has no effect. I am using a Gateway Endpoint to access S3 from inside the VPC.