rtk-ai/rtk

rtk aws eks describe-cluster compresses full cluster details to single line — critical fields silently discarded

Open

#1466 aperta il 23 apr 2026

Vedi su GitHub
 (1 commento) (0 reazioni) (0 assegnatari)Rust (2914 fork)batch import
area:clibugeffort-mediumfilter-qualityhelp wantedpriority:high

Metriche repository

Star
 (48.085 star)
Metriche merge PR
 (Merge medio 11g 1h) (45 PR mergiate in 30 g)

Descrizione

Bug

rtk aws eks describe-cluster aggressively compresses the full JSON response down to a single line containing only name, status, version, and endpoint. All other fields are silently discarded — VPC config, security groups, networking, logging, OIDC identity, certificate authority, encryption config, tags, access config, and health status are completely lost.

This makes rtk aws eks describe-cluster unusable for any inspection or debugging task. The whole point of describe-cluster is to see the detailed configuration.

Steps to reproduce

# Raw AWS output (80 lines of detailed JSON)
rtk proxy aws eks describe-cluster --name <cluster-name> --region <region>

# RTK-filtered output (1 line)
rtk aws eks describe-cluster --name <cluster-name> --region <region>

Actual rtk output:

production-cluster ACTIVE k8s/1.28 https://ABCDEF1234567890.gr7.us-east-1.eks.amazonaws.com

Expected: Compact but complete — preserving at minimum VPC config, networking, logging, encryption, tags, access config, and health issues. These are the fields users run describe-cluster to see.

Verbose flags don't helprtk aws -v and rtk aws -vvv produce identical single-line output.

Fields silently discarded

Field Why it matters
resourcesVpcConfig (subnets, SGs, endpoints) Network debugging, connectivity issues
kubernetesNetworkConfig (service CIDR) Pod/service networking
logging.clusterLogging Audit/compliance verification
identity.oidc IAM Roles for Service Accounts setup
certificateAuthority kubeconfig generation
encryptionConfig (KMS key) Security audit
tags Cost allocation, ownership
accessConfig Auth mode verification
health.issues Cluster health — arguably the most important field

Root cause

The rtk aws JSON compaction treats describe-cluster the same as list commands — extracting a summary line. But describe commands are inspection commands where the detail is the signal. The compaction should either:

  1. Exempt describe-* / get-* single-resource subcommands from aggressive compression
  2. Use a dedicated EKS describe filter that preserves important fields in compact form
  3. Respect -v / -vvv flags to progressively reveal more detail

Related

  • #1416 — same rtk aws JSON pipeline also misclassifies cluster names as date?
  • #857 — kubectl ecosystem support (kubectl path is unaffected — passthrough works fine)

Workaround

rtk proxy aws eks describe-cluster --name <cluster-name>

Environment

  • rtk 0.36.0 (Homebrew, macOS arm64)
  • Also confirmed on 0.37.1 via #1416 (same JSON compaction pipeline)
  • Darwin arm64

Guida contributor