@cryptotaxi247 / netdata
Code Commits Issues Pulls Sessions CI/CD
master
master
md 26 KB
View source Raw

vSphere Collector Parity Matrix

Status: draft baseline from the vSphere V2 migration in PR #22458 / commit 18991fd4e.

Purpose: normalize LogicMonitor, Datadog, and mirrored open-source vSphere coverage into Netdata implementation groups. Every row is classified; there are no unknown rows.

Classification values:

  • existing-default: current Netdata vSphere collector already collects it by default.
  • new-default: safe additive object-level metric group planned for default collection in this SOW.
  • opt-in: implement only behind explicit config selectors/limits because of cardinality, cost, sensitivity, or reviewer scope.
  • covered-elsewhere: Netdata already has another collector/surface for it.
  • follow-up: valid requirement, but belongs to another ingestion path or SOW.
  • non-metric-surface: valid parity surface, but must be implemented through topology output or Functions, not as metric labels.
  • out-of-scope-pr: valid parity surface intentionally excluded from this PR by user decision.
  • excluded: intentionally not supported by default because it conflicts with product direction or would duplicate a better Netdata source.

Source Evidence

Official docs checked on 2026-05-08:

  • LogicMonitor VMware vSphere Monitoring: https://www.logicmonitor.com/support/vmware-vsphere-monitoring
  • Datadog vSphere integration: https://docs.datadoghq.com/integrations/vsphere/
  • Broadcom vSphere Web Services API:
    • VirtualMachineSnapshotInfo
    • VirtualMachineSnapshotTree
    • DatastoreSummary
    • PerformanceManager
    • VirtualDisk
    • VirtualEthernetCard
    • StoragePod
    • Network
    • NetworkSummary
    • DistributedVirtualPortgroup
    • HostHostBusAdapter
    • HostScsiDisk
    • HostMultipathInfo
    • CPU, network, disk, virtual disk, storage adapter, storage path, and power performance-counter pages
  • Broadcom vSphere Automation API:
    • Cis Tagging Tag Association
  • Broadcom vSAN Management API:
    • API overview, managed objects, endpoints, and VsanPerformanceManager

Mirrored repository evidence:

Note: local mirrored repositories are shallow clones. Evidence below is snapshot-only evidence from the checked HEAD commit; it supports current-source parity comparisons, not history/blame/timeline conclusions.

  • DataDog/integrations-core @ 1befb9012c44152b0aedfb17142041bcc9c1dc61
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:91
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:115
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:175
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:262
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:326
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:375
    • vsphere/datadog_checks/vsphere/data/conf.yaml.example:386
    • vsphere/datadog_checks/vsphere/metrics.py:74
    • vsphere/datadog_checks/vsphere/metrics.py:211
    • vsphere/datadog_checks/vsphere/metrics.py:413
    • vsphere/datadog_checks/vsphere/metrics.py:497
  • influxdata/telegraf @ 5a1147f1bb725ff8fd483ea55045506aa70db191
    • plugins/inputs/vsphere/README.md:43
    • plugins/inputs/vsphere/README.md:86
    • plugins/inputs/vsphere/README.md:145
    • plugins/inputs/vsphere/README.md:173
    • plugins/inputs/vsphere/README.md:187
    • plugins/inputs/vsphere/README.md:213
    • plugins/inputs/vsphere/vsphere.go:23
    • plugins/inputs/vsphere/vsphere.go:153
    • plugins/inputs/vsphere/vsan.go:41
    • plugins/inputs/vsphere/vsan.go:115
    • plugins/inputs/vsphere/vsan.go:201
  • grafana/vmware_exporter @ 3edc42190c6709567c0465304525f42ead2ac550
    • vsphere/test_metrics.txt:1
    • vsphere/test_metrics.txt:72
    • vsphere/test_metrics.txt:80
  • elastic/beats @ 7bbe8ee6dcfbf416c53ceb7725909d37a499846c
    • metricbeat/module/vsphere/virtualmachine/virtualmachine.go:79
    • metricbeat/module/vsphere/virtualmachine/virtualmachine.go:169
    • metricbeat/module/vsphere/virtualmachine/data.go:75
    • metricbeat/module/vsphere/datastorecluster/datastorecluster.go:46
    • metricbeat/module/vsphere/network/network.go
  • zabbix/zabbix @ bbf78e24c09c90ed9d18f1570b4fd2618981d72f
    • templates/app/vmware/template_app_vmware.yaml:1132
    • templates/app/vmware/template_app_vmware.yaml:1498
    • templates/app/vmware/template_app_vmware.yaml:1532
    • templates/app/vmware/template_app_vmware.yaml:1600
    • templates/app/vmware/template_app_vmware.yaml:1633
    • templates/app/vmware/template_app_vmware.yaml:1703
    • templates/app/vmware/template_app_vmware.yaml:1778
    • templates/app/vmware/template_app_vmware.yaml:4162
    • templates/app/vmware/template_app_vmware.yaml:4418
  • newrelic/nri-vsphere @ 9366fcd3d597ae0712c94882331042d74fe38e22
    • README.md:8
    • README.md:26
    • README.md:30
    • README.md:42
    • internal/collect/vms.go:21
    • internal/collect/vms.go:22
    • internal/collect/networks.go:14
    • internal/collect/networks.go:37
    • internal/process/datacenter.go:104
    • internal/process/hosts.go:95
    • internal/process/vms.go:113
    • vsphere-performance.metrics:22
    • vsphere-performance.metrics:140
    • test-data/README.md:1
  • open-telemetry/opentelemetry-collector-contrib @ 34ed18e037dc63e41c4b4a8356d2a13d55c768f4
    • receiver/vcenterreceiver/metadata.yaml:25
    • receiver/vcenterreceiver/metadata.yaml:75
    • receiver/vcenterreceiver/metadata.yaml:276
    • receiver/vcenterreceiver/metadata.yaml:434
    • receiver/vcenterreceiver/metadata.yaml:491
    • receiver/vcenterreceiver/metadata.yaml:689
    • receiver/vcenterreceiver/metadata.yaml:791
    • receiver/vcenterreceiver/resources.go:102
    • receiver/vcenterreceiver/internal/mockserver/README.md:1
    • receiver/vcenterreceiver/internal/mockserver/responses/cluster-vsan.xml
  • grafana/alloy @ c1b740cd7fc7d2b521304ee15c9c9f61d0d5ceb0
    • internal/component/otelcol/receiver/vcenter

Matrix

ID Surface Main sources Netdata target Classification Default policy and implementation requirements
P01 VM aggregate CPU, memory, swap, disk IO, disk max latency, network traffic, packets, drops, overall alarm status, uptime Current Netdata; LogicMonitor VM performance; Datadog VM metrics; Telegraf VM metrics Existing contexts vsphere.vm_* in vsphere-v1-compatibility-manifest.md existing-default Preserve contexts, dimensions, labels, units, and sample keys exactly. Chart IDs intentionally change under framework V2 by user decision on 2026-05-08. Empty VM performance scrape results warn and continue with VM property/status metrics and later resource surfaces.
P02 ESXi host aggregate CPU, memory, swap, disk IO, disk max latency, network traffic, packets, drops, errors, overall alarm status, uptime Current Netdata; LogicMonitor host performance; Datadog host metrics; Telegraf host metrics Existing contexts vsphere.host_* existing-default Preserve current default collection. Empty host performance scrape results warn and continue with host property/status metrics and later resource surfaces.
P03 Datastore aggregate capacity/free/used/used percent, overall status, IO throughput, IOPS, latency Current Netdata; LogicMonitor datastore usage/status/throughput; Datadog datastore metrics; Telegraf datastore metrics Existing contexts vsphere.datastore_* existing-default Preserve datastore accessibility guard: capacity/free/used are trusted only when accessible.
P04 Cluster host count, CPU/memory capacity, CPU topology, DRS/HA enabled, overall status, vMotions, DRS score/balance, VM count, DRS usage summary, aggregate performance, VM operation counters Current Netdata; LogicMonitor clusters; Datadog cluster metrics; Telegraf cluster metrics; Grafana exporter cluster metrics Existing contexts vsphere.cluster_* existing-default Preserve property-vs-perf two-phase lifecycle.
P05 Resource pool CPU/memory usage, entitlement, allocation, memory breakdown, config, overall status Current Netdata; LogicMonitor resource pools; Telegraf resource pools Existing contexts vsphere.resource_pool_* existing-default Preserve current resource-pool property refresh behavior and labels.
P06 VM snapshot aggregate count, maximum snapshot age, maximum chain depth User requirement; LogicMonitor VM snapshots; Zabbix snapshot count/latest date; Elastic snapshot info; New Relic optional snapshots; Broadcom snapshot API Implemented contexts: vsphere.vm_snapshot_count, vsphere.vm_snapshot_max_age, vsphere.vm_snapshot_max_chain_depth with labels id, datacenter, cluster, host, vm new-default Object-level per VM. Emits zero for VMs with no snapshots. Does not emit snapshot name/description/ID labels by default. Unit: snapshots, seconds, snapshots. Tests: empty, sibling, nested, zero create time, old create time.
P07 VM snapshot health alerts User requirement; LogicMonitor/Zabbix snapshot alerting surfaces Implemented health templates on vsphere.vm_snapshot_max_chain_depth and vsphere.vm_snapshot_max_age new-default Warn when chain depth > 3. Critical when max age > 24h. Alert docs and metadata added with the metrics.
P08 Datastore accessible, maintenanceMode, uncommitted, multipleHostAccess Broadcom DatastoreSummary; Datadog datastore properties; LogicMonitor datastore status/usage Implemented contexts: vsphere.datastore_accessibility_status, vsphere.datastore_maintenance_status, vsphere.datastore_multiple_host_access; existing vsphere.datastore_space_usage adds uncommitted new-default Object-level per datastore. Preserves capacity/free/uncommitted guard: values are emitted as zero when inaccessible. Maintenance and multi-host access are state-set charts with unknown for omitted API values. Initial discovery now keeps inaccessible datastores as status-only resources and datastore perf scraping skips inaccessible datastores.
P09 VM power state, connection state, VMware tools running/version status, disk consolidation-needed status, configured CPU/memory/disk/NIC counts, aggregate storage usage, guest OS name LogicMonitor VM status; Datadog property metrics; Zabbix tools/status; Elastic VM summary Implemented contexts: vsphere.vm_power_state, vsphere.vm_connection_state, vsphere.vm_tools_running_status, vsphere.vm_tools_version_status, vsphere.vm_consolidation_needed, vsphere.vm_config_cpu, vsphere.vm_config_memory, vsphere.vm_config_devices, vsphere.vm_storage_usage new-default Object-level per VM. VMs returned by vSphere and kept by include selectors are discovered regardless of power state. Non-powered-on VMs are property/status/snapshot-only; real-time performance scraping skips them. No guest hostname/IP labels by default. Tools and connection values are bounded state-set dimensions. Guest OS name remains a non-default label/property candidate because it is a free-form string.
P10 Host connection state, power state, maintenance mode LogicMonitor host status; Datadog property metrics Implemented contexts: vsphere.host_power_state, vsphere.host_connection_state, vsphere.host_maintenance_status new-default Object-level per host. Hosts returned by vSphere and kept by include selectors are discovered regardless of power state. Non-powered-on hosts are property/status-only; real-time performance scraping skips them. Collector-generated ESXi vnode routing is excluded from this PR.
P11 Cluster DRS mode/vMotion rate and HA details beyond current enabled/admission-control booleans Datadog property metrics; LogicMonitor HA/admission control Implemented contexts: existing vsphere.cluster_drs_config, vsphere.cluster_ha_config; additive vsphere.cluster_drs_mode, vsphere.cluster_drs_vmotion_rate, vsphere.cluster_ha_host_monitoring, vsphere.cluster_ha_vm_monitoring, vsphere.cluster_ha_vm_component_protection new-default Object-level per cluster. Uses bounded state-set dimensions plus a numeric vMotion recommendation threshold. Does not expose free-form cluster config strings as labels.
P12 Datacenter object counts and inventory counts LogicMonitor object count/info; Datadog datacenter metrics; Telegraf datacenter controls Implemented context: vsphere.inventory_objects with datacenters, folders, clusters, hosts, VMs, datastores, and resource-pool dimensions after include filters are applied new-default Job-level aggregate metric, not a mandatory vCenter vnode. V2 uses only the static id=inventory instance label.
P13 Datastore clusters / storage pods capacity and usage LogicMonitor datastore clusters; Broadcom StoragePod; Elastic datastorecluster module Implemented optional contexts: vsphere.datastore_cluster_space_utilization, vsphere.datastore_cluster_space_usage, vsphere.datastore_cluster_storage_drs_status opt-in Default off. Enable with collect_datastore_clusters. datastore_cluster_include matches /Datacenter/DatastoreCluster, name, or managed object ID. Labels: id, datacenter, datastore_cluster.
P14 VM virtual disk capacity by disk/device LogicMonitor VM disk capacity; Broadcom VirtualDisk; Zabbix VM storage; Datadog property/perf metrics; 2026-05-22 user decision No per-virtual-disk capacity metrics in this PR out-of-scope-pr Hard-removed before merge by user decision. Existing aggregate VM disk I/O and max-latency contexts remain default-on. Reintroduce per-disk capacity only through a focused PR with a clear need and config contract.
P15 VM virtual disk performance by disk/device Broadcom disk I/O counter docs; Datadog per-instance virtualDisk.*; Telegraf virtualDisk.*; OTel VM disk metrics; New Relic performance levels; 2026-05-22 user decision No per-virtual-disk performance metrics in this PR out-of-scope-pr Hard-removed before merge by user decision. Existing aggregate VM disk I/O and max-latency contexts remain default-on. Reintroduce per-disk performance only through a focused PR with a clear need and config contract.
P16 VM network interface throughput/packets/errors/drops by NIC Broadcom latest network counters; Datadog per-instance VM net.*; Telegraf VM instance metrics; OTel VM vNIC metrics; LogicMonitor VM interface; 2026-05-22 user decision No per-VM-network-interface metrics in this PR out-of-scope-pr Hard-removed before merge by user decision. Existing aggregate VM network traffic, packets, and drops contexts remain default-on. Reintroduce per-vNIC performance only through a focused PR with a clear need and config contract.
P17 Host physical NIC metrics by NIC Broadcom latest network counters; LogicMonitor ESXi network interfaces; Telegraf host instances; Datadog per-instance host metrics; OTel host pNIC metrics; 2026-05-22 user decision No per-host-physical-NIC metrics in this PR out-of-scope-pr Hard-removed before merge by user decision because it is a high-cardinality host child-instance surface. Existing aggregate ESXi host network metrics remain default-on. Reintroduce per-pNIC performance only through a focused PR with a clear need and config contract.
P18 Host disk/LUN/device metrics by disk/device Broadcom disk counters and HostScsiDisk; LogicMonitor ESXi disks; Datadog per-instance disk metrics; Telegraf host disk metrics; OTel host disk metrics; 2026-05-22 user decision No per-host-disk/LUN/device metrics in this PR out-of-scope-pr Hard-removed before merge by user decision because it is a high-cardinality host child-instance surface. Existing aggregate ESXi host disk metrics remain default-on. Reintroduce per-device performance only through a focused PR with a clear need and config contract.
P19 Host storage adapter metrics Broadcom storage adapter counters and HostHostBusAdapter; Datadog/Telegraf/New Relic storageAdapter metrics; 2026-05-22 user decision No per-host-storage-adapter metrics in this PR out-of-scope-pr Hard-removed before merge by user decision because it is a high-cardinality host child-instance surface. Reintroduce storage-adapter performance only through a focused PR with a clear need and config contract.
P20 Host storage path metrics Broadcom storage path counters and HostMultipathInfo; Datadog/Telegraf/New Relic storagePath metrics; 2026-05-22 user decision No per-host-storage-path metrics in this PR out-of-scope-pr Hard-removed before merge by user decision because it is a high-cardinality host child-instance surface. Reintroduce storage-path performance only through a focused PR with a clear need and config contract.
P21 Host CPU core/thread/logical processor metrics Broadcom CPU counters; LogicMonitor logical processors; Telegraf host CPU instances; New Relic host CPU instance counters; OTel idle CPU metric; 2026-05-22 user decision No per-host-CPU-instance metrics in this PR out-of-scope-pr Hard-removed before merge by user decision because it is a high-cardinality host child-instance surface. Existing aggregate ESXi host CPU metrics remain default-on. Reintroduce per-CPU-instance performance only through a focused PR with a clear need and config contract.
P22 Host and VM power/energy counters Broadcom power counters; LogicMonitor ESXi power; Datadog/Telegraf/New Relic power counters; 2026-05-22 user decision Implemented contexts: vsphere.host_power_usage, vsphere.host_power_capacity_usage, vsphere.host_power_capacity_utilization, vsphere.host_energy_usage, vsphere.vm_power_usage, vsphere.vm_energy_usage implemented collect_power_metrics was removed before merge by user decision. The collector requests aggregate vSphere power.* counters with empty instance for discovered powered-on hosts and VMs when vSphere exposes those counters. No child selector is added because this emits one aggregate set per included host/VM. Host labels: id, datacenter, cluster, host, plus opt-in enrichment labels. VM labels: id, datacenter, cluster, host, vm, plus opt-in enrichment labels.
P23 Hardware health sensors: fans, power, storage, memory, processor, system sensors LogicMonitor ESXi hardware/system sensors; SNMP vmware-esx profile SNMP profile vmware-esx; optional future direct vSphere/CIM metrics covered-elsewhere Do not duplicate by default. vSphere docs point users to snmp with vmware-esx; if direct API sensors are added later, make opt-in and map against SNMP coverage.
P24 ESXi HBA/environment/hardware health via SNMP Netdata SNMP vmware-esx overlap SNMP vmware-esx profile covered-elsewhere vSphere docs mention this as complementary per-host monitoring.
P25 vCenter Server Appliance CPU, memory, disk, filesystem, services, health, VCHA, backup LogicMonitor VCSA modules Netdata vcsa collector covered-elsewhere Keep out of this vSphere collector. vSphere docs point users to vcsa for appliance health.
P26 vSAN cluster, host, and VM capacity/performance/health/events Broadcom vSAN Management API; Datadog vSAN metrics/events; Telegraf vSAN controls; OTel vSAN metrics/fixtures Implemented opt-in contexts: vsphere.vsan_cluster_space_usage, vsphere.vsan_cluster_space_utilization, vsphere.vsan_cluster_health_status, vsphere.vsan_cluster_operations, vsphere.vsan_cluster_throughput, vsphere.vsan_cluster_latency, vsphere.vsan_cluster_congestions, vsphere.vsan_host_operations, vsphere.vsan_host_throughput, vsphere.vsan_host_latency, vsphere.vsan_host_congestions, vsphere.vsan_host_cache_hit_rate, vsphere.vsan_vm_operations, vsphere.vsan_vm_throughput, vsphere.vsan_vm_latency opt-in Default off. Enable with collect_vsan. Uses vSAN Management API only for vSAN-enabled clusters that pass vsan_cluster_include; host and VM vSAN performance entities pass vsan_host_include and vsan_vm_include. Emits cluster space/health and the OTel/Datadog common cluster/host/VM vSAN performance subset. vSAN events are excluded with P27. Deeper vSAN disk-group, disk, component, CMMDS, and all Telegraf entity-type metrics remain a residual parity gap because they need explicit Netdata NIDL/context mapping and bounded config policy before implementation.
P27 vCenter and ESXi events, alarms, event filters Datadog events; LogicMonitor LogSources; New Relic optional events Future logs/events ingestion, likely OTEL/log path out-of-scope-pr User decision 2026-05-08: do not implement events in this PR. Do not overload metric collector.
P28 vSphere tags and custom attributes as labels Broadcom Tag Association API and CustomFieldsManager; Datadog tags/attributes; Telegraf custom attributes; Elastic custom fields; New Relic tags Implemented opt-in labels vsphere_tag_<sanitized_category> and vsphere_custom_attribute_<sanitized_name> opt-in Default off. tag_categories allowlists vSphere tag categories with one glob pattern per YAML list item; multiple tags in one category are sorted and joined with the pipe character. custom_attributes allowlists custom attributes with one glob pattern per YAML list item. Discovery fails open with warnings if tag/custom-attribute APIs are unavailable.
P29 Inventory paths, folder lineage, guest hostnames, guest IPs, MAC/IQN/WWN ERI-like identity LogicMonitor topology/ERI/netscan; Datadog filters; Telegraf IP addresses; OTel inventory-path resource attributes; Broadcom guest/device/storage objects No metric labels implemented in this PR out-of-scope-pr User decision 2026-05-20: remove inventory-path and VM guest hostname/IP/OS labels before merge. Remaining MAC/IQN/WWN/device identity also remains out of scope because these are sensitive multi-value identity surfaces that need a separate product decision and bounded policy.
P30 Topology edges: cluster, datastore, network, VM topology LogicMonitor topology sources; Elastic network/datastorecluster modules; OTel resource model; New Relic object fixtures; Netdata Function topology schema and SNMP topology pattern Implemented public topology:vsphere cached Function alias with required job selector non-metric-surface Uses cached discovery state and emits topology actors/links, not metric labels. Default inventory topology includes datacenters, clusters, hosts, VMs, datastores, datastore clusters, and resource pools. Network actors and host/VM network links are included only when collect_network_topology is enabled. Canonical framework registration is also available as vsphere:topology:vsphere; topology consumers should use topology:vsphere.
P31 Network and distributed virtual port group status LogicMonitor network state; Elastic network module; New Relic network object fixtures; Broadcom NetworkSummary.accessible; Broadcom DistributedVirtualPortgroup Implemented opt-in cached topology actors through collect_network_topology non-metric-surface Default off to avoid extra vCenter discovery calls for existing users. When enabled, discovers vSphere Network objects, including distributed virtual port groups returned by the Network view, and exposes cached accessible, overall_status, type, host count, VM count, and host/VM links in the topology Function. Does not create charts, metrics, or free-form network path labels.
P32 Troubleshooter and permission/readiness checks LogicMonitor troubleshooter; Datadog service checks; Netdata Function table schema Implemented vsphere:readiness cached Function with required job selector non-metric-surface Read-only cached Function reporting target/credential presence, initialized client/discovery/performance-counter state, inventory counts, optional metric/label gates, network topology gate, and cached vSAN counts. It does not expose configured URL/credentials and does not issue extra vCenter API calls. Live permission probes remain intentionally absent from this PR.
P33 VM vnodes User discussion; Netdata agent-on-VM duplication risk; V2 host-scope spec; 2026-05-20 review decision No collector-generated VM vnodes in this PR out-of-scope-pr Hard-removed before merge. Reintroduce only through a focused PR with stable VM identity, lifecycle, docs, and duplicate-node policy.
P34 ESXi vnodes User decision; V2 host-scope spec; 2026-05-20 review decision No collector-generated ESXi vnodes in this PR out-of-scope-pr Hard-removed before merge. Reintroduce only through a focused PR with stable ESXi identity and lifecycle policy.
P35 Datastore vnodes User decision and cardinality/identity review No datastore vnodes in this SOW excluded Keep datastores as metric instances, not nodes. Revisit only with product decision.
P36 TKG/Kubernetes workload metrics inside vSphere VMs Datadog TKG note; Netdata Kubernetes collectors/agents Netdata Agent and Kubernetes collectors inside guest/cluster covered-elsewhere Do not collect container/pod/node workload metrics through vSphere.

Implementation Gates Derived From Matrix

Before framework v2 migration:

  • Preserve all existing-default rows through the compatibility manifest.
  • Add no new-default row until the v2 migration passes the manifest.

For new-default rows:

  • Add one resource type per context.
  • Use only stable state dimensions and numeric dimensions.
  • Do not add sensitive labels by default.
  • Add docs, metadata, config schema, stock config, and health alert updates in the same commit group.

For opt-in rows:

  • Each row needs a config knob and selector/allowlist when the surface can emit child-instance or user-defined labels. User decision on 2026-05-20 removed all max_* knobs from this collector; selectors and allowlists are the controls for optional resource instances and user-defined labels.
  • Default must be off.
  • Tests must cover both disabled and enabled behavior.

For covered-elsewhere rows:

  • Update vSphere docs to explain the complementary collector.
  • Do not duplicate default data unless a later SOW records a product decision.

For follow-up rows:

  • No remaining row should use follow-up in this PR unless the user explicitly asks to split it again.

For non-metric-surface rows:

  • Implement through topology output or Functions only.
  • Do not add topology/resource identity as metric labels unless that label is separately covered by an opt-in label-enrichment row.

For out-of-scope-pr rows:

  • Record the user decision and do not implement in this PR.