feat(egfx): add QoE statistics accumulation on GraphicsPipelineServer#1167
Merged
Benoît Cortier (CBenoit) merged 1 commit intoMar 16, 2026
Merged
Conversation
The server currently processes QoE Frame Acknowledge PDUs and frame acknowledgments but discards all performance data after forwarding to handler callbacks. This adds a QoeCollector that automatically accumulates client decode/render timing and frame round-trip latency, exposed via qoe_snapshot() for health monitoring consumers. QoeCollector tracks: - Client decode+render time from QoE PDUs (EMA, min/max, latest) - Frame round-trip latency from FrameInfo.sent_at correlation (EMA, min/max) New public API: - QoeSnapshot struct (Clone + Debug) - GraphicsPipelineServer::qoe_snapshot() -> Option<QoeSnapshot> - GraphicsPipelineServer::reset_qoe() No changes to GraphicsPipelineHandler trait. Zero breaking changes. Part of Devolutions#1158 (Section 11: Session Health Monitoring)
Greg Lamberson (glamberson)
force-pushed
the
feat/qoe-collector
branch
from
f62761e to
99f644a
Compare
Benoît Cortier (CBenoit)
merged commit bf694c8
into
Devolutions:master
10 checks passed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Part of #1158 (Section 11: Session Health Monitoring & Stream Observability)
Summary
Motivation
The server currently processes QoE data and discards it. Ironrdp-server logs it at warn! level. Downstream consumers reimplement accumulation in their handler callbacks. This puts the accumulation where the data naturally flows (the server already processes both PDU types and owns FrameInfo.sent_at for RTT correlation).
On Linux server deployments, session health monitoring is critical: Headless RDP servers running behind tunnels, in Proxmox clusters, or as systemd services need queryable performance data for health checks and alerting. Currently there is no way to answer "what is the client decode latency?" or "what is the frame round-trip time?" without building custom accumulation on top of the handler callbacks.
Design
QoeCollector uses exponential moving average (alpha=0.1, approximately 10-sample effective window) for both decode+render time and RTT, with min/max tracking. It mirrors the FrameTracker pattern: private accumulator owned by the server, exposed via public read-only accessors.
QoeSnapshot is Clone + Debug (not Serialize) and follows the convention of existing IronRDP data types. Consumers who need serialization can trivially project from the snapshot fields.
No changes to GraphicsPipelineHandler. The existing on_frame_ack() and on_qoe_metrics() callbacks continue to work exactly as before.
Test plan