Hi, I am trying to access the underlying s3 storage of a repository using

aws --profile lakefs s3 ls s3://<bucket>/<repo> --recursive

. I have configured the

lakefs

profile with credentials from LakeFS, and the corresponding endpoint url. I also have

AuthFullAccess

,

FSFullAccess

and

RepoManagementFullAccess

permissions. However, I still get the following error

An error occurred (AccessDenied) when calling the ListObjectsV2 operation: Forbidden

Do you have an idea what I am missing?

Hi @Cristian Caloian, Two questions- 1. What do you mean when you say I have configured the

lakefs

profile with credentials from LakeFS? do you mean you use access key and secret from lakeFS or do you mean you use the same AWS credentials your lakeFS instance use? 2. About your command-

aws --profile lakefs s3 ls s3://<bucket>/<repo> --recursive

why do you use s3://<bucket>/<repo>? you should use the storage namespace you configured for your repo, e.g.-

s3://<bucket>/<prefix>

isn't it supposed to be `s3://<repo>/<branch>/<somePath>`` ?

Thank you @Idan Novogroder and @HT for your replies. I guess I am mixing things up and I would appreciate your guidance. @Idan Novogroder for (1) I configured an aws profile named

lakefs

(in

~/.aws/credentials

) with the access key and secret from LakeFS. Regarding (2) I am trying to follow this migration guide, so it might be that I understood wrong what

<s3://sourceBucket>

in Step 3 means. In my case, if I have a LakeFS repository called

foo

with the storage namespace configured as

s3://<some-bucket>/foo

. So I thought I needed to access

s3://<some-bucket>/foo

. But I guess I should do

<s3://foo/main/>

as @HT suggested, which works with the current configuration. So, if

<s3://sourceBucket>

in the migration guide refers to the actual underlying S3 bucket, then I need the AWS credentials for the AWS instance?

If you're asking if you should use AWS credentials in your

~/.aws/credentials

file, the answer is yes. Please let me know if it worked for you 🙏🏽

I guess my first question is whether, in the context of migrating a repository from one LakeFS installation to a completely different LakeFS installation, I need to copy the underlying S3 storage. If the answer is yes then I think now I understand that it needs the AWS credentials.