update

Author: StellaHuang95

analysis/kusto/00-telemetry-validation.kql

@@ -6,7 +6,7 @@
 
 
 // =============================================================================
-// CHECK 0: UniqueMachines with Telemetry Changes
+// CHECK 0: UniqueMachines with new telemetry
 // Shows unique machines per version for builds >= 1.23.10781012 (all stable + pre-release).
 // Use this to confirm the latest pre-release build has real users.
 // =============================================================================
@@ -23,9 +23,7 @@ filtered
 
 
 // =============================================================================
-// CHECK 1: Are all 4 events arriving? (3 new + 1 enhanced)
-// Expected: at least 1 row per event. If any row shows 0, that event is broken.
-// Broken down by extension version so you can confirm the new build is reporting.
+// CHECK 1: Are all 4 events arriving? (Only versions with ⚠️ NO DATA YET are displayed)
 // =============================================================================
 let allEvents = datatable(EventName: string) [
     "ms-python.vscode-python-envs/manager_registration.failed",
@@ -53,11 +51,12 @@ allEvents
     EventCount = coalesce(EventCount, 0), 
     UniqueMachines = coalesce(UniqueMachines, 0),
     Status = iff(coalesce(EventCount, 0) > 0, "✅ RECEIVING DATA", "⚠️ NO DATA YET")
+| where Status == "⚠️ NO DATA YET"
 | order by EventName asc, ExtVersion desc;
 
 
 // =============================================================================
-// CHECK 2: Enhanced event — does "result" property exist?
+// CHECK 2: MANAGER_REGISTRATION_DURATION event success rate
 // The existing event previously had only "duration". Now it should have "result".
 // Expected: rows with result = "success" (should be the majority) and maybe a few "error".
 //           If result is empty/"", the property isn't being sent correctly.
@@ -81,7 +80,7 @@ RawEventsVSCodeExt
 
 
 // =============================================================================
-// CHECK 3: MANAGER_REGISTRATION.SKIPPED — property validation
+// CHECK 3: MANAGER_REGISTRATION.SKIPPED (Percentage of users don't have each tool)
 // Expected: managerName in {conda, pyenv, pipenv, poetry}, reason = "tool_not_found"
 //           These should be common — most users don't have all 4 tools.
 // =============================================================================
@@ -100,10 +99,12 @@ RawEventsVSCodeExt
 | extend ExtVersion = tostring(Properties["common.extversion"])
 | extend _minor = toint(extract("^1\\.(\\d+)", 1, ExtVersion)), _patch = tolong(extract("^1\\.\\d+\\.(\\d+)", 1, ExtVersion))
 | where _minor > 23 or (_minor == 23 and _patch >= 10781012)
+| extend ManagerName = tostring(Properties.managername), Reason = tostring(Properties.reason)
+| where isnotempty(ManagerName) and isnotempty(Reason)
 | summarize 
     EventCount = count(),
     UniqueMachines = dcount(VSCodeMachineId)
-    by ManagerName = tostring(Properties.managername), Reason = tostring(Properties.reason)
+    by ManagerName, Reason
 | extend TotalUniqueMachines = totalMachines
 | extend MachinePct = round(todouble(UniqueMachines) / todouble(TotalUniqueMachines) * 100, 1)
 | order by EventCount desc;

analysis/kusto/01-failure-stage-breakdown.kql

@@ -0,0 +1,18 @@
+// Query 1 — Registration Failure Stage Breakdown per Manager (28 days)
+// For each manager that failed, shows WHICH stage in the registration flow broke.
+// failureStage is hierarchical: "getPipenv:nativeFinderRefresh", "constructCondaSourcingStatus", etc.
+// High counts at a specific stage → that code path is the priority fix target.
+let endDate = startofday(now()-1d);
+RawEventsVSCodeExt
+| where ServerTimestamp > endDate-28d and ServerTimestamp < endDate
+| where EventName == "ms-python.vscode-python-envs/manager_registration.failed"
+| extend ExtVersion = tostring(Properties["common.extversion"])
+| extend _minor = toint(extract("^1\\.(\\d+)", 1, ExtVersion)), _patch = tolong(extract("^1\\.\\d+\\.(\\d+)", 1, ExtVersion))
+| where _minor > 23 or (_minor == 23 and _patch >= 10781012)
+| extend ManagerName = tostring(Properties.managername)
+| extend FailureStage = tostring(Properties.failurestage)
+| summarize
+    FailureCount = count(),
+    AffectedMachines = dcount(VSCodeMachineId)
+    by ManagerName, FailureStage
+| order by ManagerName asc, AffectedMachines desc

analysis/kusto/02-error-type-x-failure-stage.kql

@@ -0,0 +1,20 @@
+// Query 2 — Error Type × Failure Stage Matrix (28 days)
+// Cross-tabulates errorType (what kind of error) with failureStage (where it happened).
+// This is the key diagnostic view: e.g., "connection_error at nativeFinderRefresh" means
+// PET process dies during native finder, while "tool_not_found at pathLookup" means the
+// tool binary wasn't on PATH. Prioritize cells with the highest AffectedMachines.
+let endDate = startofday(now()-1d);
+RawEventsVSCodeExt
+| where ServerTimestamp > endDate-28d and ServerTimestamp < endDate
+| where EventName == "ms-python.vscode-python-envs/manager_registration.failed"
+| extend ExtVersion = tostring(Properties["common.extversion"])
+| extend _minor = toint(extract("^1\\.(\\d+)", 1, ExtVersion)), _patch = tolong(extract("^1\\.\\d+\\.(\\d+)", 1, ExtVersion))
+| where _minor > 23 or (_minor == 23 and _patch >= 10781012)
+| extend ManagerName = tostring(Properties.managername)
+| extend ErrorType = tostring(Properties.errortype)
+| extend FailureStage = tostring(Properties.failurestage)
+| summarize
+    FailureCount = count(),
+    AffectedMachines = dcount(VSCodeMachineId)
+    by ManagerName, ErrorType, FailureStage
+| order by AffectedMachines desc, ManagerName asc

analysis/kusto/03-failure-stage-daily-trend.kql

@@ -0,0 +1,19 @@
+// Query 3 — Daily Registration Failures by Manager and Stage (14 days)
+// Day-by-day failure counts per manager + failureStage combination.
+// Use this to detect if a specific stage started failing more after a release.
+// A spike in one stage on a specific day → regression in that code path.
+let endDate = startofday(now()-1d);
+RawEventsVSCodeExt
+| where ServerTimestamp > endDate-14d and ServerTimestamp < endDate
+| where EventName == "ms-python.vscode-python-envs/manager_registration.failed"
+| extend ExtVersion = tostring(Properties["common.extversion"])
+| extend _minor = toint(extract("^1\\.(\\d+)", 1, ExtVersion)), _patch = tolong(extract("^1\\.\\d+\\.(\\d+)", 1, ExtVersion))
+| where _minor > 23 or (_minor == 23 and _patch >= 10781012)
+| extend Day = startofday(ServerTimestamp)
+| extend ManagerName = tostring(Properties.managername)
+| extend FailureStage = tostring(Properties.failurestage)
+| summarize
+    FailureCount = count(),
+    AffectedMachines = dcount(VSCodeMachineId)
+    by Day, ManagerName, FailureStage
+| order by Day asc, ManagerName asc, AffectedMachines desc