[XCM] Observability & Debuggability

[acatangiu]

Message Tracking

Make XCM messages easier to trace:

• XCM transport layers emit success and failure events containing local XCM message_id,
• Use SetTopic to set an unique global topic in all XCM programs built by FRAME or Polkadot pallets and/or System Chains runtimes,
• Make sure the "topic" is inherited by all onward XCM messages generated as a result of local execution of an XCM program,
• All docs, examples and tests also use SetTopic in the same way,
• Associate XCM global topic with local message_id and emit event so that indexers can quickly triage messages and categorize them into cross-chain flows.

Debuggability

• Emit error-specific events for XCM execution errors: [xcm] Investigate the possibility of ProcessXcmMessage / XcmExecutor emitting an event on failure. #2780,
• Runtime API guide/tool for replaying XCM messages with full logging, e.g. https://franciscoaguirre.github.io/polkadot-dry-runner/
• Chopsticks guide on how to replay XCMs with full logging,
• Publish log-enabled runtime versions: Publish log-enabled runtime versions #6123
• Log errors with target: xcm including error condition specifics and context,
• Add tracing logs to all SDK-provided XCM filters, helpers, matchers and types. E.g. everything in:
  • https://github.com/paritytech/polkadot-sdk/blob/master/polkadot/xcm/xcm-builder/src/
  • https://github.com/paritytech/polkadot-sdk/tree/master/cumulus/parachains/runtimes/assets/common/src
  • runtime-defined XCM filters/converters (just one example)

Developer tracing

• Enable logger for all crates unit-tests: Add a logger to unit tests #6125

---

Related sub-issues:

• Improve XCM debuggability  #644
• [XCM] don't ignore errors #2408
• messageQueue not showing Error when success is no #478

[bkontur]

* [ ]  Make sure the "topic" is inherited by all onward XCM messages generated as a result of local execution of an XCM program,

We should be able to add (or extend existing) xcm-emulated tests with SetTopic check, that initial tracking ID generated by the sender is the same on the receiver chain:

• PenpalA uses relaychain as reserve and sends to the PenpalB (or other chain)
• PenpalA uses AssetHub as reserve and sends to the PenpalB (or other chain)