Real-time monitoring dashboard for Proxmox VE, PBS, Docker, Kubernetes, TrueNAS and vSphere. Self-hosted, with alerts for failed backups, full disks, unhealthy containers and offline nodes.
Go aialertsdashboarddockerdocker-monitoringgohomelabinfrastructure-monitoringkubernetesmonitoringmspproxmoxproxmox-backup-serverproxmox-mail-gatewayproxmox-veself-hostedtruenastypescriptvspherewebhooks
Preserve cross-installation relationships in privacy-safe reports
pulse-triage[bot]7a9121481160
diff --git a/.github/ISSUE_TEMPLATE/bug_report.yml b/.github/ISSUE_TEMPLATE/bug_report.ymlindex 00ee19c599..f88b12f02a 100644--- a/.github/ISSUE_TEMPLATE/bug_report.yml+++ b/.github/ISSUE_TEMPLATE/bug_report.yml@@ -22,6 +22,9 @@ body:description: |What is broken, where do you see it, and what is the impact? Distinguish the affected target (for example, a VM, LXC, Docker container, NAS or Pulse itself) from how the Pulse server is installed.For missing readings, say whether the affected workload's usual UI was still responsive when the problem occurred. Use existing observations; write "unknown" if you cannot safely tell. Do not run another update, backup or guest-agent probe just to answer.++ For reports involving several hosts or clusters, say which nodes share a cluster, which belong to independent installations, whether node names or guest IDs (VMIDs) repeat, and whether they share a backup destination.+ Use consistent aliases for private names and addresses across the description, screenshots and logs; preserve which values repeat (for example, cluster A/node 1 and standalone B/node 1). Use existing observations, or write "unknown"; do not add, rename, remove or re-enrol anything just to answer.placeholder: A clear and concise description of the problem.validations:required: truediff --git a/.github/ISSUE_TEMPLATE/v6_rc_feedback.yml b/.github/ISSUE_TEMPLATE/v6_rc_feedback.ymlindex d6020a50a5..631a8730c7 100644--- a/.github/ISSUE_TEMPLATE/v6_rc_feedback.yml+++ b/.github/ISSUE_TEMPLATE/v6_rc_feedback.yml@@ -132,6 +132,9 @@ body:description: |Describe the problem, friction, or feedback and where you see it. Distinguish the affected target (for example, a VM, LXC, Docker container, NAS or Pulse itself) from how the Pulse server is installed.For missing readings, say whether the affected workload's usual UI was still responsive when the problem occurred. Use existing observations; write "unknown" if you cannot safely tell. Do not run another update, backup or guest-agent probe just to answer.++ For reports involving several hosts or clusters, say which nodes share a cluster, which belong to independent installations, whether node names or guest IDs (VMIDs) repeat, and whether they share a backup destination.+ Use consistent aliases for private names and addresses across the description, screenshots and logs; preserve which values repeat (for example, cluster A/node 1 and standalone B/node 1). Use existing observations, or write "unknown"; do not add, rename, remove or re-enrol anything just to answer.placeholder: Pulse upgraded cleanly, but the new storage filters are hard to understand and I could not find the previous view quickly.validations:required: truediff --git a/docs/ISSUE_TRIAGE.md b/docs/ISSUE_TRIAGE.mdindex b7a955b406..d513b690e9 100644--- a/docs/ISSUE_TRIAGE.md+++ b/docs/ISSUE_TRIAGE.md@@ -57,6 +57,24 @@ well as the original title and screenshots before choosing diagnostics. When areport shifts to a different target, retain both topics rather than assumingthe new symptom is the original defect or a duplicate.+For reports involving several hosts or clusters, preserve which nodes share a+cluster, which belong to independent installations, whether node names or guest+IDs (VMIDs) repeat, and whether they share a backup destination. Display names+and VMIDs are not globally unique; a distinct Pulse display name alone does not+prove that every reading is isolated. Use consistent aliases for private names+and addresses across the description, screenshots and logs, preserving which+values repeat: cluster A/node 1 and standalone B/node 1 are distinct targets with+the same native node name. Do not infer that a backup or agent belongs to one+installation from a matching name or VMID alone, or classify a new report as a+duplicate just because those identifiers overlap.++Use relationships already supplied in the full thread and attachments. Unknown+relationships remain unknown; do not ask for public hostnames, addresses or a+configuration dump, or add, rename, remove or re-enrol anything to obtain this+context. Existing reports need no refile. Retain every affected surface (for+example, node errors, backup status and Docker monitoring); recovery in one does+not establish recovery in the others.+Missing readings and an unresponsive workload need different investigation andrecovery paths. Use the reporter's existing observations of the workload's usualUI, not Pulse's displayed connection status, to distinguish them where possible.diff --git a/frontend-modern/public/docs/ISSUE_TRIAGE.md b/frontend-modern/public/docs/ISSUE_TRIAGE.mdindex b7a955b406..d513b690e9 100644--- a/frontend-modern/public/docs/ISSUE_TRIAGE.md+++ b/frontend-modern/public/docs/ISSUE_TRIAGE.md@@ -57,6 +57,24 @@ well as the original title and screenshots before choosing diagnostics. When areport shifts to a different target, retain both topics rather than assumingthe new symptom is the original defect or a duplicate.+For reports involving several hosts or clusters, preserve which nodes share a+cluster, which belong to independent installations, whether node names or guest+IDs (VMIDs) repeat, and whether they share a backup destination. Display names+and VMIDs are not globally unique; a distinct Pulse display name alone does not+prove that every reading is isolated. Use consistent aliases for private names+and addresses across the description, screenshots and logs, preserving which+values repeat: cluster A/node 1 and standalone B/node 1 are distinct targets with+the same native node name. Do not infer that a backup or agent belongs to one+installation from a matching name or VMID alone, or classify a new report as a+duplicate just because those identifiers overlap.++Use relationships already supplied in the full thread and attachments. Unknown+relationships remain unknown; do not ask for public hostnames, addresses or a+configuration dump, or add, rename, remove or re-enrol anything to obtain this+context. Existing reports need no refile. Retain every affected surface (for+example, node errors, backup status and Docker monitoring); recovery in one does+not establish recovery in the others.+Missing readings and an unresponsive workload need different investigation andrecovery paths. Use the reporter's existing observations of the workload's usualUI, not Pulse's displayed connection status, to distinguish them where possible.diff --git a/scripts/release_control/release_promotion_policy_test.py b/scripts/release_control/release_promotion_policy_test.pyindex e3b57cf9e6..897bb45948 100644--- a/scripts/release_control/release_promotion_policy_test.py+++ b/scripts/release_control/release_promotion_policy_test.py@@ -1910,6 +1910,58 @@ class ReleasePromotionPolicyTest(unittest.TestCase):):self.assertIn(precaution, triage)+ def test_cross_installation_intake_preserves_relationships_without_private_identity(self) -> None:+ for name, summary_id in (+ ("bug_report.yml", "bug_description"),+ ("v6_rc_feedback.yml", "summary"),+ ):+ with self.subTest(form=name):+ form = yaml.load(read(f".github/ISSUE_TEMPLATE/{name}"), Loader=UniqueKeyLoader)+ fields = {field["id"]: field for field in form["body"] if "id" in field}+ self.assertEqual(len(fields), sum("id" in field for field in form["body"]))+ summary = fields[summary_id]+ self.assertEqual(summary["type"], "textarea")+ self.assertTrue(summary["validations"]["required"])+ self.assertNotIn("render", summary["attributes"])+ prose = normalize_ws(summary["attributes"]["description"])+ for distinction in (+ "For reports involving several hosts or clusters",+ "which nodes share a cluster", "independent installations",+ "whether node names or guest IDs (VMIDs) repeat",+ "share a backup destination", "consistent aliases for private names and addresses",+ "across the description, screenshots and logs", "preserve which values repeat",+ "cluster A/node 1 and standalone B/node 1", "Use existing observations",+ 'write "unknown"', "do not add, rename, remove or re-enrol anything just to answer",+ ):+ self.assertIn(distinction, prose)+ # The conditional context belongs in the existing report, not a new+ # required identity field or a request to collect a configuration.+ self.assertNotIn("topology", fields)+ self.assertNotIn("hostname", fields)+ self.assertFalse(fields["environment"]["validations"]["required"])++ documents = ("docs/ISSUE_TRIAGE.md", "frontend-modern/public/docs/ISSUE_TRIAGE.md")+ self.assertEqual(read(documents[0]), read(documents[1]))+ for document in documents:+ with self.subTest(document=document):+ prose = normalize_ws(read(document))+ for distinction in (+ "which nodes share a cluster", "independent installations",+ "whether node names or guest IDs (VMIDs) repeat", "share a backup destination",+ "Display names and VMIDs are not globally unique",+ "distinct Pulse display name alone does not prove",+ "consistent aliases for private names and addresses",+ "preserving which values repeat", "same native node name",+ "Do not infer that a backup or agent belongs",+ "duplicate just because those identifiers overlap",+ "full thread and attachments", "Unknown relationships remain unknown",+ "do not ask for public hostnames, addresses or a configuration dump",+ "add, rename, remove or re-enrol anything", "Existing reports need no refile",+ "node errors, backup status and Docker monitoring",+ "recovery in one does not establish recovery in the others",+ ):+ self.assertIn(distinction, prose)+def test_demo_site_copy_points_at_current_release_packet_index(self) -> None:demo_copy = read("docs/releases/V6_RC_DEMO_SITE_COPY.md")self.assertIn("docs/RELEASE_NOTES.md", demo_copy)