Como usar
- Copie o comando e verifique o pacote antes de instalar no seu ambiente de agente.
- Consulte no original as ferramentas, credenciais e ambientes necessários.
- Descreva os dados de entrada e a entrega esperada; confira a saída com as fontes.
npx skills add https://github.com/dynatrace/dynatrace-for-ai --skill dt-obs-network-devicesAntes de começar
- A instalação não inclui assinatura de modelo, franquia API ou acesso aos seus arquivos.
- Estrelas e instalações indicam atividade, não um teste independente de qualidade ou segurança.
Renderizado do repositório de origem, preservando títulos, exemplos, código, tabelas, links e imagens.
Network Devices Skill
Analyze SNMP-monitored network infrastructure in Dynatrace — switches, routers, firewalls, load balancers, access points, and any other device polled by a network device extension (the SNMP-generic and vendor SNMP extensions). This skill covers the device and interface data model and routes each question to the right reference file. All DQL lives in the reference files and was validated against a live tenant with dtctl query.
What Network Device Monitoring Tells You
SNMP extensions poll network hardware and expose it in Dynatrace as three layers:
- Topology — an inventory of devices and their interfaces as Smartscape nodes (
EXT_NETWORK_DEVICE,EXT_NETWORK_INTERFACE), with attributes (vendor, model, OS/firmware, location, contact, IPs, MACs, link speed) and relationships (which interface belongs to which device; which devices are neighbors via LLDP/CDP). - Metrics — time series for device health (CPU, memory, uptime) and per-interface health (operational/admin status, in/out throughput, errors, discards, packet mix), under the
com.dynatrace.extension.network_device.*namespace. - Logs — event-driven records: SNMP traps (link-up/down, hardware faults), syslog messages (config changes, authentication, interface events), and auto-discovery activity (devices and neighbors found per poll cycle).
This answers questions such as: Which interfaces are down? Which links are saturated? Is this router's CPU pegged? Which devices rebooted? What is connected to this switch? What traps has this device sent? What syslog errors are recurring?
The Three Layers (and how they join)
| Layer | Query with | Grain | Key fields | |
|---|---|---|---|---|
| Topology | smartscapeNodes "EXT_NETWORK_DEVICE" / smartscapeNodes "EXT_NETWORK_INTERFACE" | One row per device / interface (current inventory) | id, name, monitoring_mode, device_type, location, ip, speed, edges via belongs_to / calls | |
| Metrics | timeseries … com.dynatrace.extension.network_device.* | Time series per device / interface | grouped by dt.smartscape.ext_network_device and (interface metrics) dt.smartscape.ext_network_interface | |
| Logs | `fetch logs \ | filter dt.openpipeline.source == "…"` | Individual log records (traps, syslog, discovery) | joined to topology by expanding the device ip[] array and matching device.address / dt.ingest.source.ip — see references/logs.md |
The join key is the Smartscape ID. Every metric carries a dt.smartscape.ext_network_device dimension (and interface metrics also carry dt.smartscape.ext_network_interface). Use lookup [smartscapeNodes …] to attach human-readable device/interface names to metric results, and to attach live metrics to an inventory listing. Both reference files show this pattern.
Monitoring Mode Determines What Data Exists
Every device has a monitoring_mode. This is the first thing to check — it decides whether metrics exist at all:
monitoring_mode | Meaning | Has metrics? | Has full attributes? |
|---|---|---|---|
| `Extension` | Directly polled by an SNMP extension | Yes — full com.dynatrace.extension.network_device.* set | Yes |
| `Discovery` | Discovered on the network but not directly polled | No | Partial |
| `Neighbor` | Known only because a polled device names it as an LLDP/CDP neighbor | No | Minimal (name, chassis MAC) |
If a device has no metrics, check itsmonitoring_modebefore assuming a data gap — onlyExtension-mode devices are polled. When listing "monitored devices," excludeNeighbor(and usuallyDiscovery).
Routing
- Inventory, attributes, or topology (list devices, find a device's interfaces, map neighbors, read model/firmware/location) → references/topology-model.md.
- Health or performance (CPU, memory, uptime, interface up/down, throughput, saturation, errors) → references/metrics.md.
- Event-driven signals (SNMP traps, syslog messages, discovery activity) → references/logs.md.
- Most real questions combine layers (e.g. "which interfaces on the core router are down" = metric status filtered to one device, joined to interface names). The metrics and logs references show these joined queries.
Related Skills
Network device monitoring is one part of Dynatrace network observability. Route elsewhere when the question is not about device/interface health or topology:
| The user wants… | Use |
|---|---|
| Traffic between entities, top talkers, conversations, connection health (the flows through the network) | dt-obs-network-flows — including NetFlow/IPFIX/sFlow that these same devices export |
| Host NIC throughput / packet drops on OneAgent-monitored hosts (not SNMP devices) | dt-obs-hosts → references/host-metrics.md |
| Service request rate, latency, error rate | dt-obs-services |
| Setting up or configuring an extension (this skill covers the resulting data model, not extension authoring) | dt-obs-extensions |
General Smartscape traversal syntax (traverse, smartscapeEdges, references[…]) | dt-dql-essentials |
Network devices vs. network flows. This skill answers "what is the state of the box and its ports" (SNMP device/interface health and topology).dt-obs-network-flowsanswers "what traffic is crossing the network" (conversations, bytes, talkers). The same router appears in both: here as anEXT_NETWORK_DEVICEwith interface counters, there as a NetFlow exporter. A future macro `dt-obs-network` skill may unify them; until then, cross-link.
Reference Files
- references/topology-model.md — Validated. The Smartscape model:
EXT_NETWORK_DEVICEandEXT_NETWORK_INTERFACEnode types with full field reference, monitoring modes, and thebelongs_to/callsedges. DQL for device inventory (with filtering by mode/type/location/vendor), listing a device's interfaces, resolving an interface to its device, and mapping device/interface neighbor topology. - references/metrics.md — Validated. The
com.dynatrace.extension.network_device.*metric catalog with units, dimensions, and field-typing rules. DQL for device health (top CPU, memory % with fallback, uptime and reboot detection) and interface health (down interfaces, throughput and saturation vs. link speed, error/discard counts, per-device interface detail), each joined to device/interface names. - references/logs.md — Validated. The three log sources: SNMP traps (
extension:com.dynatrace.extension.snmp-traps-generic), syslog (extension:syslog), and auto-discovery activity (extension:com.dynatrace.extension.snmp-auto-discovery). Covers the IP-to-Smartscape join pattern (device.address/dt.ingest.source.ip→snmp.ip), field references for all three sources, and DQL for recent traps with device names, trap frequency by OID, error syslog, top event patterns, combined per-device view, and LLDP/CDP neighbor discovery logs.
