charts, salt: List and check dashboard UIDs

The dashboards we provision with MetalK8s have (or should have, which is
why some of this commit's changes are setting some UIDs) a UID. Starting
with Grafana v8.0, linking dashboards using their slug is not available
anymore (see
https://grafana.com/docs/grafana/latest/release-notes/release-notes-8-0-0/#breaking-changes
).

To make sure we can use these UIDs safely in links (mostly from MetalK8s
UI), we start by listing all of them in a JSON file, and enforce that
they are correct when rendering Helm charts.

Verifying the final deployed dashboards will be done in a subsequent
commit, through a post-install E2E test.

See: #3475

charts/grafana_dashboard_uids.json

@@ -0,0 +1,33 @@
+{
+  "Alertmanager / Overview": "alertmanager-overview",
+  "Kubernetes / API server": "09ec8aa1e996d6ffcd6817bbaff4db1b",
+  "Kubernetes / Networking / Cluster": "ff635a025bcfea7bc3dd4f508990a3e9",
+  "Kubernetes / Controller Manager": "72e0e05bef5099e5f049b05fdc429ed4",
+  "etcd": "c2f4e12cdf69feb95caa41a5a1b423d9",
+  "CoreDNS": "vkQ0UHxik",
+  "Kubernetes / Compute Resources / Cluster": "efa86fd1d0c121a26444b636a3f509a8",
+  "Kubernetes / Compute Resources / Namespace (Pods)": "85a562078cdf77779eaa1add43ccec1e",
+  "Kubernetes / Compute Resources / Node (Pods)": "200ac8fdbfbb74b39aff88118e4d1c2c",
+  "Kubernetes / Compute Resources / Pod": "6581e46e4e5c7ba40a07646395ef7b23",
+  "Kubernetes / Compute Resources / Workload": "a164a7f0339f99e89cea5cb47e9be617",
+  "Kubernetes / Compute Resources / Namespace (Workloads)": "a87fb0d919ec0ea5f6543124e16c42a5",
+  "Kubernetes / Kubelet": "3138fa155d5915769fbded898ac09fd9",
+  "Kubernetes / Networking / Namespace (Pods)": "8b7a8b326d7a6f1f04244066368c67af",
+  "Kubernetes / Networking / Namespace (Workload)": "bbb2a765a623ae38130206c7d94a160f",
+  "USE Method / Cluster": "3e97d1d02672cdd0861f4c97c64f89b2",
+  "USE Method / Node": "fac67cfbe174d3ef53eb473d73d9212f",
+  "Nodes": "fa49a4706d07a042595b664c87fb33ea",
+  "Nodes (Detailed)": "node-exporter-full",
+  "Kubernetes / Persistent Volumes": "919b92a8e8041bd567af9edab12c840c",
+  "Kubernetes / Networking / Pod": "7a18067ce943a40ae25454675c19ff5c",
+  "Prometheus / Overview": "prometheus-overview",
+  "Kubernetes / Proxy": "632e265de029684c40b21cb76bca4f94",
+  "Kubernetes / Scheduler": "2e6b6a3b4bddf1427b3a55aa1311c656",
+  "Kubernetes / StatefulSets": "a31c1f46e6f727cb37c0d731a7245005",
+  "Kubernetes / Networking / Workload": "728bf77cc1166d2f3133bf25846876cc",
+  "Loki": "g6fe30815b172c9da7e813c15ddfe607",
+  "Logs": "a7e130cb82be229d6f3edbfd0a438001",
+  "Fluent Bit": "fluentbit",
+  "NGINX Ingress controller": "nginx",
+  "Request Handling Performance": "4GFbkOsZk"
+}

charts/render.py

@@ -29,6 +29,8 @@
 import argparse
 import copy
 import io
+import json
+import pathlib
 import re
 import sys
 import subprocess
@@ -210,6 +212,29 @@ def remove_prometheus_rules(template, drop_rules):
     return updated_template
 
 
+DASHBOARD_UIDS_FILE = pathlib.Path(__file__).parent / "grafana_dashboard_uids.json"
+DASHBOARD_UIDS = json.loads(DASHBOARD_UIDS_FILE.read_text())
+
+
+def patch_grafana_dashboards(manifest):
+    for fname in manifest["data"]:
+        dashboard = json.loads(manifest["data"][fname])
+        title = dashboard.get("title")
+        assert title in DASHBOARD_UIDS, f"Found unknown Grafana dashboard: '{title}'"
+        found_uid = dashboard.get("uid")
+        expected_uid = DASHBOARD_UIDS[title]
+        if found_uid:
+            assert found_uid == expected_uid, (
+                f"UID mismatch for Grafana dashboard '{title}': "
+                f"found '{found_uid}', expected '{expected_uid}'"
+            )
+        else:
+            dashboard["uid"] = expected_uid
+            manifest["data"][fname] = json.dumps(dashboard, indent=4, sort_keys=True)
+
+    return manifest
+
+
 def main():
     parser = argparse.ArgumentParser()
     parser.add_argument("name", help="Denotes the name of the chart")
@@ -307,12 +332,16 @@ def __call__(self, parser, args, values, option_string=None):
             drop_prometheus_rules = yaml.safe_load(fd)
 
     def fixup(doc):
-        if (
-            drop_prometheus_rules
-            and isinstance(doc, dict)
-            and doc.get("kind") == "PrometheusRule"
-        ):
-            doc = remove_prometheus_rules(doc, drop_prometheus_rules)
+        if isinstance(doc, dict):
+            kind = doc.get("kind")
+            if drop_prometheus_rules and kind == "PrometheusRule":
+                doc = remove_prometheus_rules(doc, drop_prometheus_rules)
+            if (
+                kind == "ConfigMap"
+                and doc.get("metadata", {}).get("labels", {}).get("grafana_dashboard")
+                == "1"
+            ):
+                doc = patch_grafana_dashboards(doc)
 
         return (
             fixup_metadata(namespace=args.namespace, doc=fixup_doc(doc=doc))