JSON view does not update content on version change

[somerandow]

Describe the bug

In 1.15.4, when viewing a secret in the the read-only JSON view, the drop-down menu to select a different version of the secret does not reload or update the content of the JSON view. This appears to occur for both complex and non-complex kv secrets.

For example, version 2 of a secret, using the JSON view:

then selecting version 1 from the drop-down, the version number and creation time change, but the content does not:

However, when viewing the secret in the "default" view, the content does change. Version 2:

and then version 1:

I believe this may be related to #24290, as I can confirm what I believe to be the correct behavior works in 1.15.2.

To Reproduce Steps to reproduce the behavior:

1. Create a new kv secret, using default or JSON editor
2. Create a second version of the secret with different contents
3. Toggle the JSON view on
4. Switch back and forth between versions using the drop-down, secret contents do not change
5. Toggle JSON view off
6. Switch back and forth between versions using drop-down, secret contents do change

Expected behavior The JSON view should update its contents to reflect the content of the version of the secret currently selected.

This expected behavior is consistent with what I'm able to test in Vault 1.15.2. Willing to provide screenshots of expected behavior if needed.

Environment:

• Vault Server Version (retrieve with vault status): 1.15.4
• Vault CLI Version (retrieve with vault version): N/A
• Server Operating System/Architecture: OpenShift (Kubernetes 1.25.12)

Vault server configuration file(s):

disable_mlock = true
ui = true
cluster_name = "XXXXXX"
listener "tcp" {
  tls_disable = 0
  address = "[::]:8200"
  cluster_address = "[::]:8201"
  http_read_timeout = "600s"
  tls_cert_file = "/tls/tls.crt"
  tls_key_file = "/tls/tls.key"
  telemetry {
    unauthenticated_metrics_access = "true"
  }
}
storage "raft" {
  path = "/vault/data"
  performance_multiplier = 1
  autopilot_reconcile_interval = "5s"
  autopilot_update_interval = "1s"
  retry_join {
    leader_tls_servername = "XXXX"
    leader_ca_cert_file = "/tls/ca.crt"
    auto_join = "provider=k8s namespace=XXX"
  }
autopilot {
    cleanup_dead_servers = "true"
    last_contact_threshold = "10s"
    dead_server_last_contact_threshold = "1h"
    max_trailing_logs = 1000
    min_quorum = 3
    server_stabilization_time = "10s"
  }
}
service_registration "kubernetes" {}
seal "gcpckms" {XXXXXXX}
telemetry {
  prometheus_retention_time = "120s",
  disable_hostname = true
}

Additional context

As a workaround, selecting the version from the drop-down, and then reloading the page does force the content to update, but it's not made clear to end-users that this workaround exists.

[Monkeychip]

@somerandow thank you for reporting this issue. I have been able to replicate and can confirm this is indeed a bug. I'll report back when a fix is in the works and what release it will be targeted towards 1.15.6 and 1.16.0 should have this fix in them.