PHP Opc Ua
Menu
Home Documentation About
Sign in
opcua-cli · master
Docs · Command · dump:nodeset

`dump:nodeset`

Export a live server's namespace to a NodeSet2.xml file. Reverse of generate:nodeset — start from the server, end with a file you can feed to the code generator.

Walk a server's address space and emit a NodeSet2.xml.

Usage

text signature 
opcua-cli dump:nodeset <endpoint> --output=<file.xml> [--namespace=<index>] [global-options]
Argument Meaning
<endpoint> The OPC UA server URL. Required.
Option Default Effect
--output=PATH required Output file path (.xml)
--namespace=N all non-zero namespaces Dump only namespace index N

Connects to the server, browses recursively, reads node attributes in batches, writes them as a NodeSet2.xml.

Examples

Dump everything

bash terminal — full dump 
opcua-cli dump:nodeset opc.tcp://plc.local:4840 --output=PLC.NodeSet2.xml

Walks every node in every non-zero namespace. On a server with 50 000 nodes this takes a few minutes; the output file is megabytes.

Dump a single namespace

bash terminal — namespace-filtered 
opcua-cli dump:nodeset opc.tcp://plc.local:4840 --output=PLC-ns2.NodeSet2.xml --namespace=2

Most production deployments care only about the vendor's custom namespace (typically 2). Filter to it for a much smaller dump.

Then generate from it

bash terminal — round-trip 
# 1. Export from the server
opcua-cli dump:nodeset opc.tcp://plc.local:4840 \
    --output=MyPLC.NodeSet2.xml --namespace=2

# 2. Generate PHP types from the dump
opcua-cli generate:nodeset MyPLC.NodeSet2.xml \
    --output=src/Generated/MyPLC/ \
    --namespace="App\\OpcUa\\MyPLC"

# 3. Use in your code (loadGeneratedTypes)

This is the canonical "server has no published NodeSet2.xml" workflow. See Code generation · Dump from a server.

What ends up in the XML

The dump captures:

  • Nodes — every node visited, with NodeId, BrowseName, DisplayName, Description, NodeClass.
  • Attributes — DataType, ValueRank, AccessLevel, Historizing, IsAbstract, etc. (per the OPC UA Attribute catalogue).
  • References — every HasComponent, HasProperty, HasTypeDefinition, Organizes, … pointing out of each node.
  • Aliases — namespace 0 well-known references referenced by the dumped nodes (e.g. HasComponent = i=47).
  • Required models — declares the spec's namespace URIs the dump depends on.

What the dump does not capture:

  • Node values. The dump records structure, not runtime values. A node's Value attribute is not dumped (it changes continuously).
  • Custom DataType definitions that the server has not itself encoded as standard DataTypeDefinition attributes. The generator can still produce NodeId constants and enums, but custom structures may need manual work.

How it interacts with auto-generation

The XML the dump produces is plain NodeSet2 — feed it to generate:nodeset and the output is regular generated PHP. This is how you get typed PHP for a server whose vendor never published a NodeSet2.xml.

How it maps to the library

You ran The CLI does
dump:nodeset <endpoint> --output=… $client->browseRecursive(...) + per-node single-attribute read() calls (DataType, ValueRank, IsAbstract, Symmetric, DataTypeDefinition) + NodeSetXmlBuilder::build(...)

The internal flow is browse-then-read-then-build. Each attribute read is a separate read() call — there is no readMulti() batch here. For very large address spaces, the per-node read step dominates the runtime.

When to use it

  • Vendor-specific PLC with no published spec. Dump, generate, use the generated PHP types in your application.
  • Reverse-engineering a new server. A grep-able XML is easier to inspect than running browse --recursive repeatedly.
  • Snapshot for testing. Dump once; replay in a test environment against a faked server that serves the same shape.
  • Documentation. The XML can be diffed across firmware revisions to track namespace changes.

What the XML is not good for

  • Live values. Use read, watch, or subscriptions.
  • Method call signatures. Methods are nodes (dumped), but their InputArguments / OutputArguments properties are ExtensionObjects — the dump captures them but they may decode incompletely.
  • As-published spec. A dump is what this specific server publishes, which may differ from the OPC Foundation's canonical spec.

Common pitfalls

  • Output file overwritten silently. No confirmation prompt. Don't dump on top of a file you wanted to keep.
  • Long-running on huge address spaces. Plan for it — a 50 000-node server takes minutes. Use --namespace=N to narrow.
  • Server times out. Some servers limit per-session operation count. If the dump fails partway, narrow the namespace filter and run multiple dumps.
  • The dump captures attribute values you might not expect. ValueRank, ArrayDimensions, AccessLevel — these are static- per-node and useful. But MinimumSamplingInterval is also captured and may surprise you.
Last updated · 2026-05-25
Edit on GitHub Report issue

← Previous

generate:nodeset

Next →

Endpoint URLs

Documentation