Closed bizcochillo closed 10 months ago
@bizcochillo Hey, going to look into this, any bits of information you can share may help me reproduce/debug.
As a first step, I'll try to create account roles and deploy a cluster using the latest versions of Terraform and RHCS (what was once called OCM) to see if I can reproduce issue.
OK first issue I hit while attempting to deploy a cluster (OIDC managed from Windows 10), local folder path I'd used was too long, I'll address the issue and retry.
│ Error: Failed to download module │ │ on ..\oidc_provider\[main.tf](http://main.tf/) line 74: │ 74: module "operator_roles_and_oidc_provider" { │ │ Could not download module "operator_roles_and_oidc_provider" (..\oidc_provider\main.tf:74) source code from │ "git::https://github.com/terraform-redhat/terraform-aws-rosa-sts?ref=v0.0.12": error downloading │ '[https://github.com/terraform-redhat/terraform-aws-rosa-sts?ref=v0.0.12'](https://github.com/terraform-redhat/terraform-aws-rosa-sts?ref=v0.0.12%27): C:\Program Files\Git\cmd\git.exe exited with 128: Cloning │ into '.terraform\modules\oidc_config.operator_roles_and_oidc_provider'... │ error: unable to open loose object ca95569d51db05b3d00055fe1360ad5670f41102: Filename too long │ error: unable to open loose object ca95569d51db05b3d00055fe1360ad5670f41102: Filename too long
Shorting path via moving git repo to c:\terraform-provider-rhcs resolved the above issue. Windows is more susceptible to longer folder paths than Linux.
From a Windows 10 system, successfully created account roles and deployed an OIDC managed cluster OCP 4.13.5 Using versions: Terraform v1.5.4 on windows_amd64
Final install lines: rhcs_cluster_wait.rosa_cluster: Creation complete after 28m14s Apply complete! Resources: 2 added, 0 changed, 0 destroyed. Outputs: cluster_id = "25ejrtekdoi0m59qq3md0um08r213kdr" ....
$ rosa describe cluster -c ts-mw0608
Name: ts-mw0608
ID: 25ejrtekdoi0m59qq3md0um08r213kdr
External ID: 2a36b35d-4493-4995-b908-628c5a5a6579
Control Plane: Customer Hosted
OpenShift Version:
Channel Group: stable
DNS: ts-mw0608.2me4.s1.devshift.org
AWS Account: 42546.......
API URL: https://api.ts-mw0608.2me4.....
Console URL: https://console-openshift-console.apps.ts-mw060......
Region: us-east-1
Multi-AZ: false
Nodes:
Hi @tshefi, thank you for your feedback.
We hit this in two Windows 10 machines and we wanted to install a 4.12.x STS ROSA cluster. We experience this when we were investigating a terraform based approach, just a couple of weeks before the renaming from terraform-redhat/ocm to terraform-redhat/rhcs (version 1.0.4, maybe).
As per the logs provided in my first message, it seems to be some kind of failure by handling the https://api.openshift.com/api/clusters_mgmt/v1/aws_inquiries/sts_credential_requests response, probably by the underlying AWS SDK, but obviously I had no more information.
I don't have those machines by now, but doing the exactly same steps from a fresh windows installation reproduced the error then. I will go through it again and will provide feedback.
With best regards
Hi @bizcochillo
Thanks for the update, waiting to hear from you. Shall the need arise I can retry again this time a 4.12.x classic STS ROSA cluster on api.openshift.com.
@bizcochillo and @tshefi Thank you for bringing and checking this issue I'm closing this issue as it old one Feel free to reopen this in case the issue gets reproduced again
Hello,
we have observed a different behaviour when fetching operator_roles data source from Windows machines. While in Linux hosts the data source data.ocm_rosa_operator_roles successfully populates the operator_roles variables, in Windows we observe an error when calling the operator list.
Debug output in Windows systems
While in Linux the operator list is retrieved correctly:
With best regards, Abel.