pcluster-manager : creation failure ?

[aws-samples]

pcluster-manager : creation failure ?

[sean-smith]

Can you share your config? I just spun up pcluster manager in ca-central-1 to check

[nyue]

I deleted the stack a few hours ago.

I will try again now and we can compare notes.

[nyue]

Still failing for me, how to retrieve the config you need ?

Resource handler returned message: "Error occurred during operation 'SSM execution 'e54a07ec-ac53-4a5e-8023-fd2ecc2ecd2d' failed for image arn: 'arn:aws:imagebuilder:ca-central-1:083230063072:image/importpublicecrimage-3-2-0-39d6cac0-bb92-11ec-8357-0234fea36e06/3.2.0/1' with status = 'TimedOut' in state = 'BUILDING' and failure message = 'SSM Agent verification failed''." (RequestToken: 206a5c68-c8bf-80ef-8a0c-e31042cab07d, HandlerErrorCode: GeneralServiceException)

[charlesg3]

Hi @nyue -- I just tried creating a new Pcluster Manager in the ca-central-1 region without issue. We haven't changed anything with the ParallelClusterApi resource, perhaps it was a transient issue or an issue with the VPC setup. Have you had success in this region before?

[nyue]

Yes, I have been using this region for over a week

There was a previous ticket I raise about creation failure which was addressed about 2 weeks ago and since than, I was able to create pcluster-manager

I have tried twice in the last 24 hours, both failed.

Cheers

[nyue]

For my situation, I have a work around for now.

The following UI which says the information is optional, in my case, I need to provide explicit (default) values, it is unable to pick up my VPC and Subnet information to build the ECRImage part

[sean-smith]

These are required if you don't have a default VPC or that VPC doesn't have internet connectivity. We previously had these as AWS intrinsic parameter types, i.e. it would autofill based on vpc's in your account, however we had to change this because it didn't allow making them optional, which imho is the default case.

[robertdeheer]

I'm seeing this issue even when the vpc_id and subnet are set.

[jpellman]

@robertdeheer - Not sure if this is still a problem for you, but I also encountered this issue when setting vpc_id and subnet and discovered that there are other conditions that need to be fulfilled for the CloudFormation template to properly deploy. Namely, the documentation here indicates that IPv4 address auto-assignment must be enabled if you're using a public subnet and that private subnets need for NAT to be configured. Network ACLs and security groups can also be problematic apparently.