AWS signing region = dummy

[tivvit]

I ran into an issue with custom s3 endpoint. The S3 SDK is using v4 signing which is using region and I need to change it to something different than dummy. Loki is using weaveworks/common where this issue was recently fixed (weaveworks/common@ecd64a5). But Loki is not using an upstream version it is using a fork https://github.com/sandlis/weaveworks-common/tree/server-listen-addr (https://github.com/grafana/loki/blob/master/Gopkg.toml#L31). That should be solved after merging this weaveworks/common#153 (#945) which is quite surprisingly a different branch then mentioned earlier but that is not important.

It would be great to rebase the fork to master or to go back to upstream (which is probably not possible because of the features added in the fork).

I will try if the code added to weaveworks/common will solve my problem but I am quite sure it will. I wanted to change the dependency but the relations of the projects are not very clear to me, so I will leave to someone who knows the project better than me. If there is anything I can help with it would be a pleasure.

[tivvit]

I confirm that the setup works for me with the code from upstream and correctly set AWS_REGION env variable.

[tivvit]

The code may be seen here tivvit@4774e7a

[ThoreKr]

This would also fix an issue (#1051) with the authentication headers being stripped from a request when connecting to a custom endpoint which redirects from http to https.

[ThoreKr]

@tivvit How do you determine whether an Endpoint should be http or https?

[tivvit]

Good question @ThoreKr

I am using http, therefore I hit the general case tivvit@4774e7a#diff-a55d7f0f417b8521b203aa3ff35c9bafR56

I suppose loki aws url may be https:// instead s3:// but maybe it would break something I did not test it.

[stale]

This issue has been automatically marked as stale because it has not had any activity in the past 30 days. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.