Error: RequestError: send request failed ... s3 ... cloud-object-storage ... dial tcp ... i/o timeout

[65156]

Executing Terraform from a CICD pipeline outside of IBM Cloud, we get the following error

is there some IBMC CLI commands we need to use to auth properly to prevent this from happening? presuming the terraform provider auth block is not enough, we are using a simple API KEY authentication method, it has full administrative rights to the entire IBMC account.

Terraform Plan Output

│ Error: RequestError: send request failed
--
2800 | │ caused by: Get "https://s3.direct.us-east.cloud-object-storage.appdomain.cloud/": dial tcp 161.26.0.31:443: i/o timeout
2801 | │
2802 | │   with ibm_cos_bucket.cos_bucket["images-sbx-east-images-sbx-bucket-east"],
2803 | │   on resource.tf line 377, in resource "ibm_cos_bucket" "cos_bucket":
2804 | │  377: resource "ibm_cos_bucket" "cos_bucket" {
2805 | │

DEBUG LOG, different resource captured but its the same error from the above.

2025-04-17T22:26:24.020Z [ERROR] provider.terraform-provider-ibm_v1.77.1: Response contains error diagnostic: @module=sdk.proto tf_proto_version=5.7 tf_provider_addr=provider tf_req_id=ebca3a14-bdc9-2c09-9a6a-8fe0be29bbf3 tf_rpc=ReadResource diagnostic_detail= diagnostic_severity=ERROR diagnostic_summary="RequestError: send request failed
--
45742 | caused by: Get "https://s3.direct.us-east.cloud-object-storage.appdomain.cloud/": dial tcp 161.26.0.31:443: i/o timeout" tf_resource_type=ibm_cos_bucket @caller=github.com/hashicorp/terraform-plugin-go@v0.25.0/tfprotov5/internal/diag/diagnostics.go:58 timestamp=2025-04-17T22:26:24.020Z
45743 | 2025-04-17T22:26:24.020Z [ERROR] vertex "ibm_cos_bucket.cos_bucket_retention[\"scc-reporting-logdna-archive-nbk3m1\"]" error: RequestError: send request failed

Similar Tickets

1. Bucket creation failure with endpoint private #4357
2. ibm_cos_bucket created with endpoint_type = "public" regardless of the actual setting #5935

[65156]

I think the method to query the resource, should not rely on a get request against the direct endpoint as this is unreachable outside the context of IBMC and ultimately the root cause of the issue, Deploying resources to IBMC via terraform should be CICD agnostic, this issue would also affect people trying to deploy direct bucket resources from Travis CICD, or Github Actions for example.

we should be able to deploy a direct or private bucket or associate to one in the code, without validating the url like how we currently are.

This should NOT be expected behavior from a CICD perspective.

[IBM-diksha]

Hi @65156 , We are looking into this issue.
Thank you.