| 1 | # vSphere Collector Parity Matrix |
| 2 | |
| 3 | Status: draft baseline from the vSphere V2 migration in PR #22458 / |
| 4 | commit `18991fd4e`. |
| 5 | |
| 6 | Purpose: normalize LogicMonitor, Datadog, and mirrored open-source vSphere |
| 7 | coverage into Netdata implementation groups. Every row is classified; there are |
| 8 | no `unknown` rows. |
| 9 | |
| 10 | Classification values: |
| 11 | |
| 12 | - `existing-default`: current Netdata vSphere collector already collects it by |
| 13 | default. |
| 14 | - `new-default`: safe additive object-level metric group planned for default |
| 15 | collection in this SOW. |
| 16 | - `opt-in`: implement only behind explicit config selectors/limits because of |
| 17 | cardinality, cost, sensitivity, or reviewer scope. |
| 18 | - `covered-elsewhere`: Netdata already has another collector/surface for it. |
| 19 | - `follow-up`: valid requirement, but belongs to another ingestion path or SOW. |
| 20 | - `non-metric-surface`: valid parity surface, but must be implemented through |
| 21 | topology output or Functions, not as metric labels. |
| 22 | - `out-of-scope-pr`: valid parity surface intentionally excluded from this PR by |
| 23 | user decision. |
| 24 | - `excluded`: intentionally not supported by default because it conflicts with |
| 25 | product direction or would duplicate a better Netdata source. |
| 26 | |
| 27 | ## Source Evidence |
| 28 | |
| 29 | Official docs checked on 2026-05-08: |
| 30 | |
| 31 | - LogicMonitor VMware vSphere Monitoring: |
| 32 | `https://www.logicmonitor.com/support/vmware-vsphere-monitoring` |
| 33 | - Datadog vSphere integration: |
| 34 | `https://docs.datadoghq.com/integrations/vsphere/` |
| 35 | - Broadcom vSphere Web Services API: |
| 36 | - `VirtualMachineSnapshotInfo` |
| 37 | - `VirtualMachineSnapshotTree` |
| 38 | - `DatastoreSummary` |
| 39 | - `PerformanceManager` |
| 40 | - `VirtualDisk` |
| 41 | - `VirtualEthernetCard` |
| 42 | - `StoragePod` |
| 43 | - `Network` |
| 44 | - `NetworkSummary` |
| 45 | - `DistributedVirtualPortgroup` |
| 46 | - `HostHostBusAdapter` |
| 47 | - `HostScsiDisk` |
| 48 | - `HostMultipathInfo` |
| 49 | - CPU, network, disk, virtual disk, storage adapter, storage path, and power |
| 50 | performance-counter pages |
| 51 | - Broadcom vSphere Automation API: |
| 52 | - `Cis Tagging Tag Association` |
| 53 | - Broadcom vSAN Management API: |
| 54 | - API overview, managed objects, endpoints, and `VsanPerformanceManager` |
| 55 | |
| 56 | Mirrored repository evidence: |
| 57 | |
| 58 | Note: local mirrored repositories are shallow clones. Evidence below is |
| 59 | snapshot-only evidence from the checked HEAD commit; it supports current-source |
| 60 | parity comparisons, not history/blame/timeline conclusions. |
| 61 | |
| 62 | - `DataDog/integrations-core @ 1befb9012c44152b0aedfb17142041bcc9c1dc61` |
| 63 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:91` |
| 64 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:115` |
| 65 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:175` |
| 66 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:262` |
| 67 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:326` |
| 68 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:375` |
| 69 | - `vsphere/datadog_checks/vsphere/data/conf.yaml.example:386` |
| 70 | - `vsphere/datadog_checks/vsphere/metrics.py:74` |
| 71 | - `vsphere/datadog_checks/vsphere/metrics.py:211` |
| 72 | - `vsphere/datadog_checks/vsphere/metrics.py:413` |
| 73 | - `vsphere/datadog_checks/vsphere/metrics.py:497` |
| 74 | - `influxdata/telegraf @ 5a1147f1bb725ff8fd483ea55045506aa70db191` |
| 75 | - `plugins/inputs/vsphere/README.md:43` |
| 76 | - `plugins/inputs/vsphere/README.md:86` |
| 77 | - `plugins/inputs/vsphere/README.md:145` |
| 78 | - `plugins/inputs/vsphere/README.md:173` |
| 79 | - `plugins/inputs/vsphere/README.md:187` |
| 80 | - `plugins/inputs/vsphere/README.md:213` |
| 81 | - `plugins/inputs/vsphere/vsphere.go:23` |
| 82 | - `plugins/inputs/vsphere/vsphere.go:153` |
| 83 | - `plugins/inputs/vsphere/vsan.go:41` |
| 84 | - `plugins/inputs/vsphere/vsan.go:115` |
| 85 | - `plugins/inputs/vsphere/vsan.go:201` |
| 86 | - `grafana/vmware_exporter @ 3edc42190c6709567c0465304525f42ead2ac550` |
| 87 | - `vsphere/test_metrics.txt:1` |
| 88 | - `vsphere/test_metrics.txt:72` |
| 89 | - `vsphere/test_metrics.txt:80` |
| 90 | - `elastic/beats @ 7bbe8ee6dcfbf416c53ceb7725909d37a499846c` |
| 91 | - `metricbeat/module/vsphere/virtualmachine/virtualmachine.go:79` |
| 92 | - `metricbeat/module/vsphere/virtualmachine/virtualmachine.go:169` |
| 93 | - `metricbeat/module/vsphere/virtualmachine/data.go:75` |
| 94 | - `metricbeat/module/vsphere/datastorecluster/datastorecluster.go:46` |
| 95 | - `metricbeat/module/vsphere/network/network.go` |
| 96 | - `zabbix/zabbix @ bbf78e24c09c90ed9d18f1570b4fd2618981d72f` |
| 97 | - `templates/app/vmware/template_app_vmware.yaml:1132` |
| 98 | - `templates/app/vmware/template_app_vmware.yaml:1498` |
| 99 | - `templates/app/vmware/template_app_vmware.yaml:1532` |
| 100 | - `templates/app/vmware/template_app_vmware.yaml:1600` |
| 101 | - `templates/app/vmware/template_app_vmware.yaml:1633` |
| 102 | - `templates/app/vmware/template_app_vmware.yaml:1703` |
| 103 | - `templates/app/vmware/template_app_vmware.yaml:1778` |
| 104 | - `templates/app/vmware/template_app_vmware.yaml:4162` |
| 105 | - `templates/app/vmware/template_app_vmware.yaml:4418` |
| 106 | - `newrelic/nri-vsphere @ 9366fcd3d597ae0712c94882331042d74fe38e22` |
| 107 | - `README.md:8` |
| 108 | - `README.md:26` |
| 109 | - `README.md:30` |
| 110 | - `README.md:42` |
| 111 | - `internal/collect/vms.go:21` |
| 112 | - `internal/collect/vms.go:22` |
| 113 | - `internal/collect/networks.go:14` |
| 114 | - `internal/collect/networks.go:37` |
| 115 | - `internal/process/datacenter.go:104` |
| 116 | - `internal/process/hosts.go:95` |
| 117 | - `internal/process/vms.go:113` |
| 118 | - `vsphere-performance.metrics:22` |
| 119 | - `vsphere-performance.metrics:140` |
| 120 | - `test-data/README.md:1` |
| 121 | - `open-telemetry/opentelemetry-collector-contrib @ 34ed18e037dc63e41c4b4a8356d2a13d55c768f4` |
| 122 | - `receiver/vcenterreceiver/metadata.yaml:25` |
| 123 | - `receiver/vcenterreceiver/metadata.yaml:75` |
| 124 | - `receiver/vcenterreceiver/metadata.yaml:276` |
| 125 | - `receiver/vcenterreceiver/metadata.yaml:434` |
| 126 | - `receiver/vcenterreceiver/metadata.yaml:491` |
| 127 | - `receiver/vcenterreceiver/metadata.yaml:689` |
| 128 | - `receiver/vcenterreceiver/metadata.yaml:791` |
| 129 | - `receiver/vcenterreceiver/resources.go:102` |
| 130 | - `receiver/vcenterreceiver/internal/mockserver/README.md:1` |
| 131 | - `receiver/vcenterreceiver/internal/mockserver/responses/cluster-vsan.xml` |
| 132 | - `grafana/alloy @ c1b740cd7fc7d2b521304ee15c9c9f61d0d5ceb0` |
| 133 | - `internal/component/otelcol/receiver/vcenter` |
| 134 | |
| 135 | ## Matrix |
| 136 | |
| 137 | | ID | Surface | Main sources | Netdata target | Classification | Default policy and implementation requirements | |
| 138 | |---|---|---|---|---|---| |
| 139 | | 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. | |
| 140 | | 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. | |
| 141 | | 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. | |
| 142 | | 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. | |
| 143 | | 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. | |
| 144 | | 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. | |
| 145 | | 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. | |
| 146 | | 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. | |
| 147 | | 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. | |
| 148 | | 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. | |
| 149 | | 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. | |
| 150 | | 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. | |
| 151 | | 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`. | |
| 152 | | 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. | |
| 153 | | 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. | |
| 154 | | 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. | |
| 155 | | 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. | |
| 156 | | 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. | |
| 157 | | 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. | |
| 158 | | 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. | |
| 159 | | 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. | |
| 160 | | 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. | |
| 161 | | 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. | |
| 162 | | 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. | |
| 163 | | 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. | |
| 164 | | 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. | |
| 165 | | 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. | |
| 166 | | 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. | |
| 167 | | 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. | |
| 168 | | 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`. | |
| 169 | | 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. | |
| 170 | | 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. | |
| 171 | | 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. | |
| 172 | | 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. | |
| 173 | | 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. | |
| 174 | | 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. | |
| 175 | |
| 176 | ## Implementation Gates Derived From Matrix |
| 177 | |
| 178 | Before framework v2 migration: |
| 179 | |
| 180 | - Preserve all `existing-default` rows through the compatibility manifest. |
| 181 | - Add no `new-default` row until the v2 migration passes the manifest. |
| 182 | |
| 183 | For `new-default` rows: |
| 184 | |
| 185 | - Add one resource type per context. |
| 186 | - Use only stable state dimensions and numeric dimensions. |
| 187 | - Do not add sensitive labels by default. |
| 188 | - Add docs, metadata, config schema, stock config, and health alert updates in |
| 189 | the same commit group. |
| 190 | |
| 191 | For `opt-in` rows: |
| 192 | |
| 193 | - Each row needs a config knob and selector/allowlist when the surface can |
| 194 | emit child-instance or user-defined labels. User decision on 2026-05-20 |
| 195 | removed all `max_*` knobs from this collector; selectors and allowlists are |
| 196 | the controls for optional resource instances and user-defined labels. |
| 197 | - Default must be off. |
| 198 | - Tests must cover both disabled and enabled behavior. |
| 199 | |
| 200 | For `covered-elsewhere` rows: |
| 201 | |
| 202 | - Update vSphere docs to explain the complementary collector. |
| 203 | - Do not duplicate default data unless a later SOW records a product decision. |
| 204 | |
| 205 | For `follow-up` rows: |
| 206 | |
| 207 | - No remaining row should use `follow-up` in this PR unless the user explicitly |
| 208 | asks to split it again. |
| 209 | |
| 210 | For `non-metric-surface` rows: |
| 211 | |
| 212 | - Implement through topology output or Functions only. |
| 213 | - Do not add topology/resource identity as metric labels unless that label is |
| 214 | separately covered by an opt-in label-enrichment row. |
| 215 | |
| 216 | For `out-of-scope-pr` rows: |
| 217 | |
| 218 | - Record the user decision and do not implement in this PR. |