Microsoft Defender for Identity tells you when its own deployment is unhealthy through health issues: a sensor that stopped talking, an expired account password, auditing that's missing on the domain. This quick reference lists every health issue Microsoft documents, with its severity, the tab it appears on, the sensor version it applies to, the usual cause and the fix that makes it close.
Codes in this quick reference
Network configuration mismatch for sensors running on VMware: VM network settings affect the sensorDefender portal (Sensor health issues)A domain controller is unreachable by a sensor: the sensor can't open LDAP connections to itDefender portal (Sensor health issues)All/Some of the selected capture network adapters on the Defender for Identity sensor are disabled or disconnectedDefender portal (Sensor health issues)Low success rate of active name resolution: IP-to-name resolution fails over 90% of the timeDefender portal (Sensor/Global health issues)No network traffic received from domain controller: port mirroring missing or brokenDefender portal (Sensor/Global health issues)Read-only user password to expire shortly: Directory Service account password expires within 30 daysDefender portal (Global health issues)Read-only user password expired: sensors stop running and collect no new dataDefender portal (Global health issues)Sensor outdated (v2): the sensor version can't communicate with the cloud serviceDefender portal (Sensor/Global health issues)
+ 14 more codes in the tables below.
How to read these codes#
Open the Microsoft Defender portal and go to Settings › Identities › Health issues (under Deployment). Issues sit on two tabs: Global health issues for domain-level or aggregated problems, and Sensor health issues for a single sensor. Filter by status, name or severity; select an issue for details.
- Status: an issue is Open until Defender for Identity detects that the cause is gone and closes it automatically. Admins with the Azure ATP (workspace name) Administrators role can also close an issue, or suppress it for seven days during planned work such as taking a domain controller offline.
- Daily checks: several configuration checks (auditing, power mode) run once a day per sensor or per domain, so allow up to a day for those to close.
- Sensor version: many issues apply only to the classic sensor (2.x). Sensor v3.x issues are marked 3.x.
- Email alerts: Settings › Identities › Health issues notifications emails recipients when an issue opens (sensor 2.x).
To list open issues across all sensors, call Microsoft Graph (permission SecurityIdentitiesHealth.Read.All):
Connect-MgGraph -Scopes SecurityIdentitiesHealth.Read.All
$uri = "https://graph.microsoft.com/v1.0/security/identities/healthIssues?`$filter=Status eq 'open'"
(Invoke-MgGraphRequest -Method GET -Uri $uri).value | ForEach-Object {
[pscustomobject]@{
Issue = $_.displayName
Severity = $_.severity
Type = $_.healthIssueType
Sensors = $_.sensorDNSNames -join ', '
}
} | Format-Table -AutoSizeIssues already covered step by step in Defender for Identity sensors: installation and health issues are included below for completeness.
Sensor service and version#
| Health issue | Severity, tab, sensor | Usual cause | Fix |
|---|---|---|---|
| Sensor stopped communicating | Medium, Sensor, all | No data or keep-alive for longer than allowed (5 minutes by default): network block, or a long restart | Check routers, firewalls and proxies between the sensor and the cloud service |
| Sensor service failed to start | High, Sensor, all | Service down for at least 30 minutes | Read the sensor logs for the root cause (see "How to verify a fix") |
| Sensor outdated (v2) | Medium, both, 2.x | Version too old to talk to the cloud service | Update manually and find out why auto-update failed; reinstall from the latest package if needed |
| Sensor outdated (v3) | Medium, both, 3.x | Server lacks the required Windows cumulative update | Install the latest cumulative update, then check the version on the Sensors page |
| Sensor running on an operating system that will soon become unsupported | Medium, Sensor, 2.x | Older Windows Server; 2012 and 2012 R2 ended support on October 10, 2023 | Upgrade the server operating system |
| Sensor running on an unsupported operating system | High, Sensor, all | As above, already unsupported | Upgrade the server operating system |
| Sensor failed to write to the custom log path | Low, Sensor, 2.x | The SensorCustomLogLocation path can't be created | Stop AATPSensorUpdater and AATPSensor, fix the path (or set it to null) in the sensor configuration file, start both |
Capacity and packet capture#
| Health issue | Severity, tab, sensor | Usual cause | Fix |
|---|---|---|---|
| Sensor reached a memory resource limit | Medium, Sensor, 2.x | High memory use on the domain controller; the sensor restarts itself and data is only partly monitored | Add RAM, or add domain controllers to the site to spread the load |
| Some Windows events are not being analyzed | Medium, both, 2.x | More events than the sensor can process | Add CPU and memory; on a standalone sensor, forward only required events or split them |
| Some network traffic could not be analyzed | Medium, both, 2.x | More traffic than the sensor can process | Add CPU and memory; standalone: monitor fewer DCs. On VMware, set Large Send Offload V2 (IPv4) and IPv4 TSO Offload to 0 or Disabled inside Windows |
| Some ETW events are not being analyzed | Medium, both, 2.x | More ETW events than the sensor can process | Add CPU and memory |
| Power mode isn't configured for optimal processor performance | Low, Sensor, 2.x | Power plan other than High Performance | Set High Performance, or minimum and maximum processor state to 100 |
| Sensor has issues with packet capturing component | High, Sensor, 2.x | WinPcap in use instead of Npcap | Install Npcap as Microsoft describes (the installer has included Npcap 1.0 OEM since version 2.184) |
| Sensor has issues with packet capturing component | Medium, Sensor, 2.x | Npcap older than 1.0, the minimum supported | Upgrade Npcap |
| Sensor has issues with packet capturing component | High, Sensor, 2.x | Npcap installed without the required options | Reinstall Npcap with the required options |
| All/Some of the selected capture network adapters on the Defender for Identity sensor are disabled or disconnected | Medium, Sensor, 2.x | Capture adapters disabled or unplugged | Enable and connect them |
| No network traffic received from domain controller | Medium, both, 2.x | Port mirroring to a standalone sensor not configured or not working | Fix port mirroring; disable Receive Segment Coalescing (IPv4 and IPv6) on the capture adapter |
| Network configuration mismatch for sensors running on VMware | High, Sensor, 2.x | Virtual network adapter settings don't match requirements | Review adapter settings, including disabling Large Send Offload (LSO) |
Directory access and name resolution#
| Health issue | Severity, tab, sensor | Usual cause | Fix |
|---|---|---|---|
| Directory services user credentials are incorrect | Medium, Global, all | Wrong account name, password or domain; gMSA prerequisites missing; brief alerts during gMSA password rotation | Correct the Directory Service account entry. With both sensor versions, DSA credentials are still validated; remove the DSA once every sensor is v3.x |
| Read-only user password to expire shortly | Medium, Global, 2.x | Directory Service account password expires in under 30 days | Change the password, then update it in the Directory Service account settings |
| Read-only user password expired | High, Global, 2.x | Password expired; sensors stop running and collect no new data | Same fix, urgently |
| A domain controller is unreachable by a sensor | Medium, Sensor, 2.x | The sensor can't open LDAP connections to the configured DC | Check the DC is up and LDAP is allowed; configure a Directory Service account for every forest |
| Low success rate of active name resolution | Low, both, 2.x | Over 90% of lookups fail via NTLM over RPC, NetBIOS or reverse DNS | Allow inbound port 135 and 137 from sensors to all computers; enable reverse lookup zones |
| Sensor failed to retrieve Microsoft Entra Connect service configuration | Medium, Sensor, 2.x | The sensor can't read the ADSync configuration | Make sure the Microsoft Azure AD Sync service runs and grant the sensor access to the ADSync database |
Auditing and integrations#
| Health issue | Severity, tab, sensor | Usual cause | Fix |
|---|---|---|---|
| NTLM Auditing is not enabled | Medium, Sensor, all | Event 8004 isn't audited (checked daily) | On domain controllers only, set the three Network security: Restrict NTLM audit policies (below) |
| Directory Services Advanced Auditing is not enabled as required | Medium, Sensor, all | Required advanced audit subcategories missing (checked daily) | Configure them for Success and Failure, or use Set-MDIConfiguration |
| Directory Services Object Auditing is not enabled as required | Medium, Global, all | Domain SACL lacks the required object types (checked daily per domain) | Add Everyone audit entries for descendant User, Group, Computer and managed service account objects |
| Auditing on the Configuration container is not enabled as required | Medium, Global, all | Configuration partition not audited; needed where Exchange is or was installed | Configure it once per forest in ADSI Edit; it can show for several domains |
| Auditing on the ADFS container is not enabled as required | Medium, Global, all | No SACL on the AD FS configuration container | Audit Read and Write all properties for Everyone on Program Data › Microsoft › ADFS |
| Auditing for AD CS servers isn't enabled as required | Medium, Sensor, 2.x | Audit Certification Services policy or the CA audit filter missing | Enable the policy (Success and Failure), set the CA audit filter, restart certsvc |
| Sensor v3.x RPC Audit Misconfigured | Medium, Sensor, 3.x | Unified Sensor RPC Audit configuration tag missing or not applied | Apply the tag, or install the July 2026 or later cumulative update (sensor 3.0.8 or later audits RPC automatically) |
| Radius accounting (VPN integration) data ingestion failures | Low, Health issues page, 2.x | Shared secret doesn't match the VPN server | Re-enter the matching shared secret in the VPN integration settings |
Note: With sensor v3.x, turn on Settings › Identities › Advanced features › Automatic Windows auditing configuration. It applies the required auditing every 24 hours, and Microsoft notes that some v3.x environments with manual Group Policy or PowerShell auditing keep raising auditing issues even when the settings are correct; automatic auditing resolves them.
For sensor 2.x, the DefenderForIdentity PowerShell module checks and applies auditing in one pass. Run it with Domain Admin rights:
Import-Module DefenderForIdentity
New-MDIConfigurationReport -Path "C:\Reports" -Mode Domain -OpenHtmlReport # what's missing
Set-MDIConfiguration -Mode Domain -Configuration All # creates and links GPOs
auditpol.exe /get /category:* # on a DC, after gpupdate
# On the certificate authority: full CA auditing
certutil -setreg CA\AuditFilter 127
Restart-Service certsvcSet-MDIConfiguration creates and links Group Policy objects; add -CreateGpoDisabled or -SkipGpoLink to review them first. For NTLM auditing on domain controllers, set Outgoing NTLM traffic to remote servers to Audit all, Audit NTLM authentication in this domain to Enable all, and Audit Incoming NTLM Traffic to Enable auditing for all accounts.
How to verify a fix#
- Watch the issue's status: it should move to Closed on its own. Allow up to a day for issues checked daily.
- For service problems on a 2.x sensor, confirm both services run, then read the newest errors:
Get-Service AATPSensor, AATPSensorUpdater | Select-Object Name, Status, StartType
$logs = Get-ChildItem 'C:\Program Files\Azure Advanced Threat Protection Sensor' -Directory |
Sort-Object LastWriteTime -Descending | Select-Object -First 1
Get-Content (Join-Path $logs.FullName 'Logs\Microsoft.Tri.Sensor-Errors.log') -Tail 30- For auditing fixes, confirm the new events (such as 4662, 4776 and 8004) appear in the domain controller's logs, and rerun
New-MDIConfigurationReport: nothing should be flagged. - Rerun the Graph query above. An empty result means no open health issues.
Errors during installation are in the deployment logs in the installing user's temp folder; Microsoft.Tri.Sensor.Updater-Errors.log covers failed automatic updates.

