Cluster taking too long to apply a modification.

When an AWS cluster gets stuck in the "Modifying" state for an extended period, it can be frustrating, especially when it's a new cluster with no data or traffic.

This issue can occur for various reasons:

1. Background processes that might be taking longer than expected
2. Resource constraints in the AWS infrastructure
3. Network issues
4. Potential bugs in the service

Since you've already tried restarting the nodes without success, here are some steps you can take:

1. Check any available logs or CloudWatch metrics for your cluster to identify potential issues or error messages.

2. If possible, verify the health status of your cluster using service-specific endpoints (like _cluster/health for OpenSearch/Elasticsearch).

3. For reporting potential bugs without a support plan, you have a few options:

• Post on AWS re:Post (formerly AWS Forums) where AWS employees sometimes monitor and respond
• Submit feedback through the AWS Management Console (look for feedback links)
• If you're using a service with open-source components (like OpenSearch), you might be able to file an issue in their GitHub repository

4. As a last resort, if this is a critical issue and you need immediate assistance, you might consider signing up for a one-month support plan to get the issue resolved, then cancel afterward.

Unfortunately, without a support plan, direct access to AWS support for bug reporting is limited. AWS generally encourages customers to use their support plans for reporting and resolving service issues.
Sources
Elasticache stuck "modifying" when performing a failover | AWS re:Post
AWS Opensearch cluster is stuck in "Modifying" with the update | AWS re:Post
OpenSearch cluster says in domain status "Modifying" for week | AWS re:Post

Based on my experience managing AWS environments, I can understand your frustration, especially when the issue seems tied to a simple operation like updating a user password and ACL on a fresh cluster.

When a cluster stays in a "modifying" state for an unusually long time, particularly one with no data or traffic, it’s often a sign of an underlying issue, possibly even a platform-side bug. I’ve come across similar situations where certain control plane operations hang due to internal inconsistencies, and in such cases, restarting nodes rarely has any effect, just as you’ve noticed.

Unfortunately, without an AWS Support plan, there’s no guaranteed direct line to report these types of issues. However, what I typically recommend in such scenarios is to use the AWS Developer Forums, AWS re:Post, or even raise awareness via social media (e.g., tagging AWS Support on Twitter/X). Sometimes, if the issue is widespread or has been publicly flagged, AWS engineers will respond or escalate it internally.

That said, I also agree with your point that requiring a support plan just to report potential bugs feels limiting, especially for experienced users who’ve managed infrastructure independently for years. It would be ideal if AWS offered a free, minimal channel specifically for bug reports or platform issues.

In the meantime, if the cluster is new and not in production, it might be faster to delete and recreate it rather than wait indefinitely. But I’d still recommend posting the issue in a public AWS forum to help others and possibly get it flagged internally.