The e2e-aws-csi job has been consistently failing since January 16, 2026 when PR #70150 added the aws-deprovision-verification post-step: https://prow.ci.openshift.org/job-history/gs/test-platform-results/pr-logs/directory/pull-ci-openshift-origin-main-e2e-aws-csi

The step fails with:

/bin/bash: line 245: /usr/bin/jq: Argument list too long

error: failed to execute wrapped command: exit status 126

Root Cause

Script: ci-operator/step-registry/aws/deprovision/verification/aws-deprovision-verification-commands.sh:242-244

The script passes large JSON data as command-line arguments to jq, exceeding the OS ARG_MAX limit (~2MB) when the CI AWS account contains many tagged resources:

TAGGED_RESOURCES=$(jq -n --argjson results "$(printf '%s\n' "${TAGGED_RESOURCES_LIST[@]}" | jq -s '.')" \
    '{ResourceTagMappingList: ($results | map(.ResourceTagMappingList) | flatten | unique_by(.ResourceARN))}')

Required Actions

1. Fix ARG_MAX issue - Already being addressed in PR #73720, which modifies the script to use stdin/file-based input instead of command-line arguments

2. Investigate potential AWS resource leaks - The large number of tagged resources causing the ARG_MAX error may indicate that previous test runs are not properly cleaning up AWS resources in the CI account