[observability] Observability Coverage Report - 2026-08-17 #53529
Closed
Replies: 1 comment
|
This discussion has been marked as outdated by Daily Observability Report for AWF Firewall and MCP Gateway. A newer discussion is available at Discussion #53859. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Executive Summary
I analyzed a capped sample of 20 completed workflow runs from the last 7 days using the repository-wide
agenticworkflows logspull and the artifacts saved under/tmp/gh-aw/aw-mcp/logs/. All 20 sampled runs were firewall-enabled and MCP-enabled, and every one of them retained the required debugging artifacts:access.logfor the AWF firewall and MCP telemetry viarpc-messages.jsonl.There were no critical observability gaps in the sampled runs. Coverage was 100% for both firewall access logs and MCP telemetry presence. The main issue is log quality rather than log absence: 19 of 20 firewall-enabled runs did not show both allowed and blocked requests in
access.log, so blocked-path visibility remains under-exercised in most successful runs. MCP telemetry was consistently present, but it came entirely from the canonical fallbackrpc-messages.jsonl; the preferredgateway.jsonlformat was absent from this sample.Across the sample, the firewall logs contained 3587 entries with 91 blocked requests (2.5%) across 23 unique domains. MCP telemetry contained 298 JSONL events, 105 outgoing
tools/callrequests, and 8 error/difc-filtered events (2.7%).Key Alerts and Anomalies
No critical issues detected.
🔴 Critical Issues:
access.logwas present in all 20 sampled firewall-enabled runs, but only 1 run showed both allowed and blocked traffic. 18 runs showed only allowed traffic, andDaily BYOK Ollama Testrun32076040028showed only blocked traffic.rpc-messages.jsonl; the preferredgateway.jsonlformat did not appear in this sample, so duration metrics were not available from the richer source.safeoutputs.jsonlwas present in 17/20 runs; the missing 3 samples did not affect firewall or MCP coverage, but they reduce structured safe-output traceability.Coverage Summary
access.log)gateway.jsonlorrpc-messages.jsonl)📋 Detailed Run Analysis
Firewall-Enabled Runs
Missing Firewall Logs (
access.log)None in the sampled runs.
MCP-Enabled Runs
rpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlrpc-messages.jsonlMissing MCP Telemetry (no
gateway.jsonlorrpc-messages.jsonl)None in the sampled runs.
🔍 Telemetry Quality Analysis
Firewall Log Quality
access.logentries analyzed: 3587Smoke Copilotrun32080621476)Gateway Log Quality
rpc-messages.jsonlrather thangateway.jsonlAdditional Telemetry Coverage
agent-stdio.logpresent: 20/20awf-config.jsonoraw_info.jsonpresent: 20/20safeoutputs.jsonlpresent: 17/20Healthy Runs Summary
Smoke Copilotrun32080621476had complete firewall and MCP coverage with both allowed and blocked firewall traffic visible.Recommended Actions
access.log, not just allowed traffic.gateway.jsonlalongsiderpc-messages.jsonlin the MCP gateway path so duration and status metrics are available without reconstructing them from raw RPC traffic.safeoutputs.jsonlpublication across all agentic runs so structured output traces are consistently available during postmortems.📊 Historical Trends
Historical trend data was not loaded for this run. This report is a point-in-time sample of the last 7 days, capped at 20 completed runs.
Report generated automatically by the Daily Observability Report workflow
Analysis window: Last 7 days | Runs analyzed: 20
All reactions