Error: Could not connect to the endpoint URL while deploying Tutorial Model

[md_tcon07]

Hello Experts,

I am new to AI Core and followed the Tutorial Series to get to know the environment. (Make Predictions for House Prices with SAP AI Core | SAP Tutorials)

Unfortunately I've encountered an error I am unable to resolve. In the last steps of the Tutorial, while trying to deploy the model on a server, the error "Could not connect to the endpoint URL" appears.

I would be very grateful for your assistance.

[md_tcon07]

For further information: I meticulously followed every step of the Tutorial Series, i didn't change the code from the tutorial except the placeholders that were required to be replaced and the generated model is stored in my S3 Bucket. The Connection to the Bucket is stored in the Launchpad as is the connection to the GitHub. For further information feel free to ask!

[gianfranco-salomone]

Tutorial says

endpoint = "s3-eu-central-1.amazonaws.com", # Change this

It should be:

endpoint = "s3.eu-central-1.amazonaws.com", # `s3-` changed to `s3.`

Source: https://docs.aws.amazon.com/general/latest/gr/s3.html

[RjainCap68]

@Gianfranco-salomone : Thank you. Your solution works, I was able to proceed. 
@MichOnSAP : If you are still struggling, I suggest take a look at this solution.

[anurag_batra · Accepted solution]

Hi Duschl,

The botocore.exceptions.EndpointConnectionError error occurs when there is a problem with the connection to the AWS service endpoint. This can be due to a variety of reasons incorrect configuration, or the service not being available in the specified region.

Can you validate if your stored S3 object store Creds are the same as your current S3 creds?

[md_tcon07 · Accepted solution]

Hello Anurag,
we were able to fix this problem. While the S3 credentials worked for storing the ai model, for some reason we had to change them for using them to deply a model.

Thank you for answering and offering your help! 🙂

[MichOnSAP · Accepted solution]

Hello. I am in the same place but months later. I have 2 Object Store secrets. Name are: default example-dataset/house-price-toy/model mys3 example-dataset/house-price-toy Url in error seems to be correct URL: "https://s3-us-east-1.amazonaws.com/hello-aicore-data?prefix=example-dataset%2Fhouse-price-toy%2Fmode..." I redid them and both appears to work for other steps. Thanks, Mich

[RjainCap68 · Accepted solution]

@md_tcon07: can you please share what changes you did for the S3 credentials? was there any access issue for your account in S3? If so, what additional roles are needed?

[MichOnSAP]

Good afternoon RjainCap,

 I think I have the same case.

Attaching view of last run .  ( I ran it at least 5 times then gave up and posted) 

Thanks,

Mich 

d

[MichOnSAP]

Hello RjainCap,

Good question,
Yes , Below is a sample listing view of runs and one model.

Thank you ,
Mich

aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/
PRE data/
PRE model/

aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/model
PRE model/

aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/model/
PRE e2d8d89238047e54/
PRE e5040e89387a330a/
PRE eaf87aeb56cb9cd7/
PRE ebee5c01cc324e12/
PRE eed8ca84c3282328/
PRE efb703b375141f45/

aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/model/e2d8d89238047e54/
PRE housemodel/

aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/model/e2d8d89238047e54/housemodel/
2024-02-07 13:45:59 3348 model.pkl

[RjainCap68]

@MichOnSAP : Yes, even for me the model gets created and gets stored in this location. But somehow, during server deployment, this endpoint is not reachable. Apologies to ask again, but were you able to generate prediction URL, meaning the server deplyed successfully?

[MichOnSAP]

Good morning md_tcon07,

Here below is my set up and no I did not have to adjust any AWS roles.
It just comes up when it wants to publish the model back. As when it gets stuck
it will continue to publish and use the data . (xyz is just my method to mask the real key)
Thanks,
Mich

Name - default
Type - S3
Path Prefix: example-dataset/house-price-toy/model
Bucket: hello-aicore-data
Endpoint: s3-us-east-1.amazonaws.com
Region : us-east-1

verify SSL : off
Use HTTPS : on
Secret:
{
"AWS_ACCESS_KEY_ID": "xyz5",
"AWS_SECRET_ACCESS_KEY": "xyzncdl"
}

Name - mys3
Type - S3
Path Prefix: example-dataset/house-price-toy
Bucket: hello-aicore-data
Endpoint: s3-us-east-1.amazonaws.com
Region : us-east-1

verify SSL : off
Use HTTPS : on
Secret:
{
"AWS_ACCESS_KEY_ID": "xyzU5",
"AWS_SECRET_ACCESS_KEY": "xyzcdl"
}

From a local comand prompt listing ( same key to access)
aws s3 ls s3://hello-aicore-data/example-dataset/house-price-toy/
PRE data/
PRE model/

[RjainCap68]

@MichOnSAP : Are you able to deploy the server? I have same configuration but the error persists. The model gets created and gets stored in the bucket, but error occurs only during the server deployment.