Zum Hauptinhalt springen

Windows Event Log Review β€” May 2026

πŸ€– Generated with Claude Code

⚠️ Not yet reviewed

Co-Authored-By: Claude (noreply@anthropic.com)

Point-in-time review of the Windows Event Log collection set up per windows-event-log-baseline.md and azure-arc-onboarding-plan.md. Checks what is actually deployed in Azure against the documented baseline, and reviews the events that arrived.

The baseline filters themselves are deployed correctly. The problems are in coverage (three machines mis- or un-configured), signal-to-noise (one server produces ~78 % of all events), and the fact that nothing alerts on the collected data.

Scope & Method​

SubscriptionTogether-Prod (1c60fabf-272f-4ece-8b91-cf1596406d30)
WorkspaceTogether-Plattform-LogAnalyticsworkspace (9ee01003-d32e-4110-ab57-1f09be580fc3), RG LogAnalytics-Prod
Window7 days ending 2026-05-27
Servers in scopeTISWEB*, TISAPP*, TISDB*, plus TISADFS001 and TISWAP001

Method: enumerated Arc machines and their DCR associations via ARM, dumped the deployed DCR definitions, and queried the Event, W3CIISLog and Heartbeat tables. All counts below are for the 7-day window.

Findings Summary​

#FindingSeverity
C1TISWAP001 ships its events to the Fachtest workspace, not ProdπŸ”΄ High
C2tisweb02 collects no IIS logs (only tisweb01 sends W3CIISLog)πŸ”΄ High
C3TIS-OMDS is Arc-onboarded but collects no Windows event logs🟑 Medium
C4DB DCR collects the whole Application log, not SQL-restricted as documented🟑 Medium
C5No alert rules exist on any of the collected event data🟑 Medium
C6AMA agent version drift (TISAPP03 & tisweb01 on 1.36, rest on 1.42)🟒 Low
L1tisweb01 Schannel TLS failures β€” 24 417 events, ~78 % of all ingestionπŸ”΄ High
L2TISAPP03 eSign API unhandled exceptions (671) + DeveloperExceptionPage in prodπŸ”΄ High
L3TISDB02/TISDB01 failed logons β€” sqlservice via sqlservr.exe (1 540)🟑 Medium
L4ADFS: KDFv2 not enabled (185), proxy-trust auth failure (276)🟑 Medium
L5OMDS service errors on TISAPP03 / TISDB01🟒 Low

1. Coverage & Configuration​

Deployed vs. documented β€” DCR associations​

ServerRoleEvent-log DCR actually associatedOK?
tisweb01Webdcr-tis-prod-eventlogs-web (+ IISLogs-Prod, VMInsights)βœ…
tisweb02Webdcr-tis-prod-eventlogs-web β€” no IIS DCR⚠️ C2
TISAPP01Appdcr-tis-prod-eventlogs-appβœ…
tisapp02Appdcr-tis-prod-eventlogs-appβœ…
TISAPP03Appdcr-tis-prod-eventlogs-appβœ…
TISDB01DBdcr-tis-prod-eventlogs-dbβœ…
TISDB02DBdcr-tis-prod-eventlogs-dbβœ…
TISADFS001ADFSdcr-tis-prod-eventlogs-adfsβœ…
TISWAP001WAPdcr-tis-fachtest-eventlogs-web (Fachtest sub)❌ C1
TIS-OMDSβ€”none (VMInsights default only)❌ C3

The web/app/adfs DCR XPath filters match the baseline exactly. Only the DB DCR deviates (C4).

C1 β€” TISWAP001 β†’ Fachtest workspace πŸ”΄β€‹

TISWAP001's event-log association points to /subscriptions/87fbfaf0-9e66-41e0-a265-6667e057d8a3/resourceGroups/LogAnalytics-Fachtest/.../dcr-tis-fachtest-eventlogs-web β€” the Fachtest subscription. So the Prod, internet-facing ADFS Web Application Proxy sends its Windows events (including security events) into the test workspace. It is therefore absent from the Prod workspace entirely (no heartbeat, no events). The onboarding plan claims it has the Prod web DCR; reality differs.

Corroborated by L4: TISADFS001 logged event 276 ("federation server proxy was not able to authenticate to the Federation Service"), so the proxy trust itself also needs a look.

C2 β€” tisweb02 has no IIS logs πŸ”΄β€‹

W3CIISLog over 7 days: 4 669 262 rows from tisweb01, 0 from tisweb02. Only tisweb01 carries the IISLogs-Prod association. If the two web servers are load-balanced, roughly half of all HTTP access logging is missing from Azure.

C3 β€” TIS-OMDS collects no event logs πŸŸ‘β€‹

TIS-OMDS is Arc-onboarded but has only the default VMInsights DCR β€” no dcr-tis-prod-eventlogs-*. It is also not listed in the onboarding plan. Either apply the appropriate role DCR or record it as deliberately out of scope.

C4 β€” DB DCR deviates from the baseline πŸŸ‘β€‹

The baseline restricts the DB Application log to SQL providers:

Application!*[System[(Level=1 or Level=2 or Level=3) and (Provider[@Name='MSSQLSERVER'] or Provider[@Name='SQLSERVERAGENT'])]]

The deployed dcr-tis-prod-eventlogs-db collects the whole Application log instead:

Application!*[System[(Level=1 or Level=2 or Level=3)]]

Consequence: non-SQL sources (e.g. OmdsxManager, see L5) are collected from the DB servers. Either align the DCR to the baseline or update the baseline to match the decision.

C5 β€” No alerting on event data πŸŸ‘β€‹

Existing scheduled-query / metric alerts cover disk, CPU, memory and application availability only. Nothing fires on the collected events β€” audit-log cleared (1102), account lockout (4740), app-pool / .NET crash (WAS 5011 / 1026), ADFS cert expiry (364), or logon-failure spikes. The data is collected but no one is notified. See Β§4.

C6 β€” AMA agent version drift πŸŸ’β€‹

TISAPP03 and tisweb01 run AMA 1.36.0.0; the others were upgraded to 1.42.0.0 around 2026-05-22. All 8 in-scope servers report a current heartbeat; this is housekeeping only.

Doc name mismatches: onboarding plan lists TISWAP01/TISADFS01; actual Arc resource names are TISWAP001/TISADFS001.

2. Issues in the Logs (7 days)​

L1 β€” tisweb01 Schannel TLS storm πŸ”΄β€‹

SourceEventIDCountMeaning
Schannel3687421 561TLS 1.2 request with no mutually-supported cipher suite
Schannel368712 856Fatal error creating a TLS client credential, internal error state 10013

24 417 events β€” ~78 % of the 31 369 total Event rows for the whole fleet β€” and steady day over day. Both messages report SSPI process SYSTEM (PID 4) and client credential creation, i.e. outbound TLS originating from a system component on web01. It is specific to web01 (tisweb02 β‰ˆ 1 700 System errors total). This is a genuine TLS/cipher configuration fault (likely over-hardened Schannel or a scheduled task/agent negotiating an unsupported suite), not merely noise β€” and it currently buries every other System/Error signal.

L2 β€” TISAPP03 eSign API exceptions πŸ”΄β€‹

671 Γ— .NET Runtime / 1000 in 7 days (~96/day). Sample:

Category: Microsoft.AspNetCore.Diagnostics.DeveloperExceptionPageMiddleware
RequestPath: /esign.api/namirial/callback/draft
An unhandled exception has occurred while executing the request.
Exception: tis.gf.esign.api.Services.SignatureTransacti…

Two issues: (a) the eSign Namirial callback throws unhandled exceptions at a steady rate; (b) it is logged via DeveloperExceptionPageMiddleware, which should not be enabled in production β€” it returns stack traces and exception detail to callers.

L3 β€” DB failed logons (sqlservice) πŸŸ‘β€‹

Event | EventLog == "Security" | EventID == 4625: TISDB02 1 203, TISDB01 337, tisweb01 190. On the DB servers every failure is account TOGETHER\sqlservice, Logon Type 3, caller sqlservr.exe, source local (status 0xC00002FD). A recurring SQL-Server-internal authorization failure β€” low risk but high volume, and the dominant Security-log source. Worth a DBA's review (Kerberos/SPN or permission check). No 1102, 4740 or 4719 occurred in the window.

L4 β€” ADFS admin events πŸŸ‘β€‹

EventIDLevelCountMeaning
185Error7KDFv2 feature not enabled on the AD FS farm (MS security-hardening recommendation)
174Warning7RP-trust monitoring: HTTP-Artifact binding ignored β€” benign
276Error1Federation server proxy failed to authenticate to the Federation Service β€” ties to C1

Note: the ADFS DCR targets EventIDs 364/168/1021, but the events that actually occur are 185/174/276. They are caught only by the broad AD FS/Admin Level 1–3 filter β€” good that it is present; the targeted ID list could be revised.

L5 β€” OMDS service errors πŸŸ’β€‹

tis.volltextsuche.omds.import.service on TISAPP03 (37Γ—20102, 37Γ—10100, 5Γ—crash) and OmdsxManager on TISDB01 (14, plus 20102/10100). Recurring but low volume; functional, not a logging problem.

3. Recommendations​

Coverage (do first):

  1. Re-associate TISWAP001 to a Prod DCR (dcr-tis-prod-eventlogs-web, or the ADFS DCR given its role) and remove the Fachtest association.
  2. Add the IIS-log DCR association to tisweb02 so W3C logs match tisweb01.
  3. Decide on TIS-OMDS β€” apply a role DCR or document as out of scope.
  4. Align the DB DCR to the SQL-provider restriction (C4), or update the baseline to match.

Signal, not just less volume: 5. Fix web01's TLS config (L1) rather than filtering the symptom; the noise disappears with the root cause. If interim relief is needed, sample 36874 instead of dropping it. 6. Disable DeveloperExceptionPage for the eSign API in prod (L2) and triage the Namirial callback exception. 7. Revise the ADFS targeted EventID list to the IDs that actually fire (185, 276); keep the broad AD FS/Admin filter.

Detection: add the log alerts in Β§4 (none exist today, C5).

Housekeeping: upgrade AMA on TISAPP03 & tisweb01 (C6); fix the TISWAP01/TISADFS01 name mismatches and record the C3/C4 decisions in the onboarding plan / baseline.

4. Proposed Alert Rules (draft)​

Status: proposed β€” not yet created. Log-search alerts on the Event table in the Prod workspace. Scope: …/resourceGroups/loganalytics-prod/providers/Microsoft.OperationalInsights/workspaces/Together-Plattform-LogAnalyticsworkspace Action group: TIS-PROD-OnPremises_Server_Alert_Test-WE-AG (RG TIS-PROD-OnPremises_Servers-WE-RG) β€” the one the existing disk/CPU alerts already use.

AlertKQL (where-clause on Event)Window / freqThresholdSeverity
Audit log clearedEventLog=="Security" and EventID==11025 min / 5 min> 0Sev1
Account lockoutEventLog=="Security" and EventID==474015 min / 15 min> 0Sev2
Audit policy changedEventLog=="Security" and EventID==471915 min / 15 min> 0Sev2
App-pool crashSource=="Microsoft-Windows-WAS" and EventID==501115 min / 15 min> 0Sev2
.NET process crashSource==".NET Runtime" and EventID==102615 min / 15 min> 0Sev2
ADFS cert expiryEventLog=="AD FS/Admin" and EventID==3646 h / 6 h> 0Sev2
ADFS token issuance failedEventLog=="AD FS Auditing" and EventID==12021 h / 1 h> 10Sev3
Failed-logon spikeEventLog=="Security" and EventID==46251 h / 1 h> 200 per ComputerSev3
eSign exception rateSource==".NET Runtime" and EventID==1000 and Computer startswith "TISAPP"1 h / 1 h> 50Sev3
Server stopped loggingHeartbeat β€” no rows per Computer30 min / 30 min0Sev2

Example (az monitor scheduled-query create), audit-log-cleared:

WS="/subscriptions/1c60fabf-272f-4ece-8b91-cf1596406d30/resourceGroups/loganalytics-prod/providers/Microsoft.OperationalInsights/workspaces/Together-Plattform-LogAnalyticsworkspace"
AG="/subscriptions/1c60fabf-272f-4ece-8b91-cf1596406d30/resourceGroups/TIS-PROD-OnPremises_Servers-WE-RG/providers/microsoft.insights/actionGroups/TIS-PROD-OnPremises_Server_Alert_Test-WE-AG"

az monitor scheduled-query create \
--name "TIS-Prod-Security-AuditLogCleared-1102" \
--resource-group "LogAnalytics-Prod" \
--scopes "$WS" \
--action-groups "$AG" \
--condition "count 'Q' > 0" \
--condition-query Q='Event | where EventLog == "Security" and EventID == 1102' \
--evaluation-frequency 5m --window-size 5m --severity 1 \
--description "Security audit log was cleared (EventID 1102)"

Thresholds for the rate-based alerts (4625, 1000) should be set after a short baselining period β€” see the volumes in the appendix.

Appendix β€” Data Snapshot​

Event volume by computer / log / level (7 days), notable rows:

ComputerEventLogLevelCount
tisweb01SystemError24 253
tisweb02SystemError1 705
TISDB02Security(4625)1 203
TISADFS001Security(4624)1 077
TISAPP03ApplicationError756
TISDB01Security(4625)337
TISADFS001SystemError291

Queries used (run via az monitor log-analytics query --workspace 9ee01003-d32e-4110-ab57-1f09be580fc3):

// Volume by computer / log / level
Event | where TimeGenerated > ago(7d)
| summarize count() by Computer, EventLog, EventLevelName | sort by count_ desc

// What drives the System errors on web01
Event | where TimeGenerated > ago(7d) and Computer startswith "tisweb01"
and EventLog == "System" and EventLevelName == "Error"
| summarize count() by Source, EventID | sort by count_ desc

// Security events by computer / id
Event | where TimeGenerated > ago(7d) and EventLog == "Security"
| summarize count() by Computer, EventID | sort by count_ desc

// IIS coverage
W3CIISLog | where TimeGenerated > ago(7d) | summarize count() by Computer