Network Management

Cisco CPUHOG tracebacks after upgrading to SolarWinds Platform 2026.2

After upgrading to SolarWinds Observability Self-Hosted (OSH) or Network Performance Monitor (NPM) 2026.2, Cisco IOS-XE fixed/stackable switches may generate %EVENTLIB-3-CPUHOG tracebacks on the hman (Hardware Manager) process. This is caused by the new CISCO-ENTITY-FRU-CONTROL-MIB polling introduced in 2026.2. The permanent fix is available in version 2026.2.2 via a new "Enable Cisco FRU Extended Polling" toggle.

First published date

7/16/2026 11:51 PM

Last published date

7/30/2026 3:15 PM

Overview

Environment

  • SolarWinds Platform 2026.2.x (OSH or NPM)

  • Cisco IOS-XE fixed/stackable switches (e.g., Catalyst C9200, C9200R, Catalyst 3850)

  • Hardware Health SNMP polling enabled

Symptom

After upgrading to 2026.2, Cisco IOS-XE switches generate recurring syslog messages:

%EVENTLIB-3-CPUHOG: Switch 1 R0/0: hman: undefined: <duration>ms

Where <duration> exceeds 1000ms. These tracebacks may trigger user-configured syslog-based alerts.

The tracebacks:

  • Start immediately after the 2026.2 upgrade

  • Occur on Cisco IOS-XE fixed/stackable platforms (e.g., C9200R, Catalyst 3850)

  • Are typically absent on chassis/modular platforms (e.g., ASR, Nexus 9504, Catalyst 6840) where hman is designed for heavier FRU workloads

  • Do not cause traffic loss or functional impact on fixed switches — this is a process scheduling warning

Product section

Network Performance Monitor

Cause

Cause

In 2026.2, SolarWinds introduced polling of three new tables from the CISCO-ENTITY-FRU-CONTROL-MIB:

  • CiscoFRUModuleStatusTable (cefcModuleAdminStatus, cefcModuleOperStatus, cefcModuleResetReason, cefcModuleStatusLastChangeTime, cefcModuleUpTime)

  • CiscoFRUPowerStatusTable

  • CiscoFRUNotificationSettings (cefcMIBEnableStatusNotification OID 1.3.6.1.4.1.9.9.117.1.3.1, cefcEnablePSOutputChangeNotif OID 1.3.6.1.4.1.9.9.117.1.3.2)

Prior to 2026.2, these tables were not polled. The additional SNMP requests are processed by the hman (Hardware Manager) process on the switch. On fixed/stackable IOS-XE platforms, hman is not accustomed to this level of FRU SNMP processing, and exceeds the 1000ms CPUHOG threshold.

These FRU tables are polled in addition to the Preferred Cisco MIB setting (CISCO-ENTITY-SENSOR-MIB or CISCO-ENVMON-MIB). Changing the Preferred Cisco MIB does not stop FRU polling.

What Does NOT Resolve the CPUHOG

The following workarounds address the FRU Warning/alarm issue (separate from the CPUHOG) but do not stop the underlying FRU SNMP polling that causes the tracebacks:

  • Disabling FRU sensors via Manage Hardware Sensors — suppresses sensor display and events only. The Collector still polls the FRU OIDs every cycle.

  • Changing Preferred Cisco MIB to CISCO-ENVMON-MIB — only toggles between CISCO-ENTITY-SENSOR-MIB and CISCO-ENVMON-MIB for temperature/voltage/fan sensors. FRU tables are polled in addition to the preferred MIB regardless of this setting.

  • IgnoredEventTypes advanced setting — suppresses hardware health event generation only. Polling continues.

  • SQL UPDATE HWH_HardwareItem SET IsDisabled = 1 — disables individual sensor records. Polling continues.

For the FRU Warning/alarm issue specifically, kindly refer to KB article Cisco Hardware Health Sensors Show Warning State After Upgrading to OSH/NPM 2026.2.

Resolution

Resolution 1 — Upgrade to 2026.2.2 (Recommended)

Version 2026.2.2 introduces a new setting, HardwareHealth.EnableCiscoFRUExtendedPolling, that controls whether FRU module and notification sensors are polled.

Upgrade behavior:

  • For upgrades from 2026.2.x: the setting defaults to unchecked (disabled). FRU polling stops automatically after the upgrade — no manual action required.

  • For fresh installs of 2026.2.2: the setting defaults to checked (enabled).

Global toggle (post-upgrade verification):

  1. Log in to the SolarWinds Platform Web Console.

  2. Go to Admin > Polling Settings.

  3. Verify the checkbox "Enable Cisco FRU Extended Polling" is unchecked.

  4. If checked, uncheck it and click Save.

  5. Allow one polling cycle for the change to take effect.

Per-node override:

  1. Go to Settings > Manage Nodes and locate the Cisco node.

  2. Click Edit Properties.

  3. Scroll to the Hardware Health Polling section.

  4. Set "Enable polling of Cisco FRU modules and notification settings" to Disabled.

  5. Click Submit.

  6. Run Poll Now or wait for the next polling cycle.

Resolution 2 — Workaround for 2026.2.x (Before Upgrading)

If upgrading is not immediately possible, the CPUHOG tracebacks can be suppressed at the syslog processing level:

  1. Edit the syslog processing rule to add a condition: Message does not contain CPUHOG.

  2. This prevents SolarWinds from processing CPUHOG syslog messages and stops any syslog-based alerts from triggering.

What this workaround does:

  • Stops CPUHOG alert notifications from firing

  • No configuration changes required on the switches or SolarWinds polling settings

What this workaround does not do:

  • Does not stop the FRU SNMP polling from occurring

  • Does not stop the hman process from exceeding the 1000ms threshold on the switch

  • The CPUHOG syslog message will still be generated by the switch, but SolarWinds will not process or alert on it