Apache Ignite 2

Thin client lifecycle, connection and request events

Public product properties reconstructed from test evidence.

Thin client lifecycle, connection and request events

This page is a public product-spec reconstruction from Apache Ignite 2 test sources at revision 486a6367610c4c891b729e89373695ef83165345.

Thin-client event tests describe observability hooks available to applications around client lifecycle, connection handshakes/loss, and request execution.

User-facing guarantees

Traceable requirements

Only negative or ambiguous test status is annotated after a test name, for example status=failed, status=flaky, status=ignored, or status=mixed.

THIN-EVENTS-LIFECYCLE-01 — client lifecycle listeners observe start, close and startup failure

Requirement: Thin client lifecycle listeners must receive client-start, explicit-close, and startup-failure events.

Why: Applications and frameworks need lifecycle hooks for resource tracking and startup diagnostics.

Boundary: Mixed-strength evidence from direct listener assertions; does not prove listener ordering across concurrent clients.

Evidence tests:

THIN-EVENTS-CONNECTION-01 — connection listeners observe handshake and loss

Requirement: Connection listeners must report handshake start, handshake success/failure, and established connection loss/close with connection metadata.

Why: Operators need visibility into protocol negotiation and broken connections without instrumenting internal networking.

Boundary: Evidence; protocol-failure and connection-loss paths are selected fixtures, not every retry/failover path.

Evidence tests:

THIN-EVENTS-REQUEST-01 — request listeners observe request success and failure

Requirement: Request listeners must report request start and success/failure outcomes for tested thin-client query calls.

Why: Monitoring code needs to correlate API calls with low-level request traffic and failures.

Boundary: Weaker where request-id correlation and metadata assertions are weak; event presence and throwable linkage are stronger than full correlation semantics.

Evidence tests:

Important boundaries