Cisco Device Health

Cisco IOS-XE and NX-OS device health check and triage procedure. Use when troubleshooting Cisco routers, switches, or Nexus platforms — assessing CPU, memory, interfaces, routing, and environment. Covers both IOS-XE (ISR, ASR, Catalyst 9K) and NX-OS (Nexus 3K/5K/7K/9K) with platform-specific commands, thresholds, and decision trees that account for IOS-XE QFP/RP architecture and NX-OS VDC isolation.

Cisco IOS-XE / NX-OS Device Health Check

Structured triage procedure for assessing Cisco device health across IOS-XE and NX-OS platforms. Produces a prioritized findings report with severity classifications and recommended actions.

Commands are labeled [IOS-XE] or [NX-OS] where they diverge. Unlabeled commands work on both platforms.

When to Use

• Device reported as slow, dropping traffic, or unresponsive
• Scheduled health audit of IOS-XE routers/switches or Nexus platforms
• Post-change verification after upgrades, patches, or configuration changes
• Capacity planning data collection for CPU, memory, and link utilization
• Incident response when a Cisco device is suspected as the fault domain
• NX-OS VDC isolation troubleshooting — health check scoped to a specific VDC

Prerequisites

• SSH or console access to the device (privilege 1 minimum for IOS-XE, show role for NX-OS)
• IOS-XE 16.x/17.x or NX-OS 9.x/10.x (commands validated against IOS-XE 17.3+ and NX-OS 10.2+)
• Network reachability to management interface confirmed
• Awareness of the device's normal baseline (CPU, memory, traffic patterns)
• For NX-OS: identify which VDC you are operating in (show vdc current)

Procedure

Follow this sequence. Each step produces data for the final report. If the device is unresponsive, skip to Step 6 for crash recovery.

Step 1: Establish Baseline Context

Collect identity, uptime, and platform context.

[IOS-XE]

text

show version | include uptime|Version|bytes of memory
show inventory | include PID
show clock

[NX-OS]

text

show version | include uptime|system version|Memory
show inventory | include PID
show clock
show vdc current

Record: hostname, OS version, uptime, hardware model, current time. Short uptime means a recent reload or crash — investigate immediately. On NX-OS, confirm VDC context: all subsequent commands apply to this VDC only.

Step 2: CPU Utilization

IOS-XE separates Route Processor (RP) and QuantumFlow Processor (QFP). Triage both planes independently. NX-OS uses a unified supervisor model — show system resources is the primary view.

[IOS-XE]

text

show processes cpu sorted | head 20
show processes cpu history
show platform software status control-processor brief
show platform hardware qfp active statistics drop

[NX-OS]

text

show system resources
show processes cpu sort | head 20
show module

IOS-XE interpretation: Compare 5-second, 1-minute, and 5-minute RP averages. If RP CPU is normal but QFP drops are climbing, the data plane is overloaded while control plane is healthy — different remediation than RP overload. Key RP processes: IP Input (traffic storms), BGP Router (route churn), Crypto IKMP (VPN storms), IOSD (control plane congestion).

NX-OS interpretation: show system resources reports CPU as kernel + user percentages. Normal idle should be above 60%. NX-OS control plane CPU runs hotter than IOS-XE RP because the supervisor handles both planes. Key processes: netstack (forwarding), bgp (route processing), stp (spanning tree), vpc (vPC keepalive).

Step 3: Memory Utilization

[IOS-XE]

text

show memory statistics
show processes memory sorted | head 15

[NX-OS]

text

show system resources
show processes memory | head 15

IOS-XE: Calculate used percentage from processor pool: (Total - Free) / Total * 100. Check fragmentation: if largest free block < 10% of total free, memory is fragmented.

NX-OS: show system resources reports memory as total/used/free in KB. NX-OS reserves a larger kernel memory footprint than IOS-XE — used percentages above 60% are typical. Watch for MemFree dropping below 15% of total.

Step 4: Interface Health

text

show interfaces summary
show interfaces counters errors

[IOS-XE]

text

show interfaces | include line protocol|drops|error|CRC
show controllers [name]

[NX-OS]

text

show interface [name] | include errors|drops|CRC
show hardware internal errors module [n]

For each interface with errors:

• Calculate error rate: errors / (input + output packets) * 100
• CRC errors → Layer 1 issue (cabling, optics, SFP seating)
• Input errors without CRC → buffer overruns
• Output drops → congestion; check QoS policy or link capacity
• On NX-OS, show hardware internal errors reveals ASIC-level drops not visible in standard interface counters

Step 5: Routing Protocol Health

text

show ip route summary

[IOS-XE]

text

show ip bgp summary
show ip ospf neighbor
show ip eigrp neighbors

[NX-OS]

text

show bgp ipv4 unicast summary
show ip ospf neighbors
show ip eigrp neighbors
show vpc brief

Verify: expected route count, all neighbors in established/full state, no unexpected withdrawals. On NX-OS, vPC peer-link state is critical — a failed peer-link orphans traffic. BGP prefix counts that deviate >10% from baseline indicate route churn.

Step 6: Environment and Platform

[IOS-XE]

text

show environment all
show platform software status control-processor brief
show logging | include %|Error|Warning|traceback | tail 50
dir crashinfo:

[NX-OS]

text

show environment
show module
show logging last 50
show cores

Check: power supply status, fan health, temperature readings. Any environmental alarm escalates immediately. On IOS-XE, review crashinfo: for recent crash dumps. On NX-OS, show cores lists supervisor and module core dumps — presence indicates a process crash requiring investigation.

Threshold Tables

Reference: references/threshold-tables.md for detailed per-parameter thresholds.

Parameter	Normal	Warning	Critical	Platform Notes
CPU 5-min avg	< 40%	40–70%	> 70%	NX-OS: expect 5–10% higher baseline
CPU 5-sec spike	< 80%	80–90%	> 90%	IOS-XE RP only; NX-OS: use system resources
QFP drops/sec	< 100	100–1000	> 1000	IOS-XE only
Memory used	< 70%	70–85%	> 85%	NX-OS: normal baseline is 55–65%
Interface error rate	< 0.01%	0.01–0.1%	> 0.1%	Both platforms
Output drops/hr	< 100	100–1000	> 1000	Both platforms
Routing neighbors	All established	Flapping	Down	NX-OS: include vPC state
Temperature	Within spec	5°C of max	At/above max	Per-sensor spec varies

Decision Trees

Primary Triage

text

Is the device reachable?
├── No → Check console, power, environment. Collect crashinfo after recovery.
└── Yes
    ├── Identify platform → IOS-XE or NX-OS?
    │
    ├── [IOS-XE] CPU issue?
    │   ├── RP CPU high, QFP normal → Control plane overload
    │   │   ├── BGP Router high → Route churn, check peer stability
    │   │   ├── IP Input high → Traffic storm, check ACLs/CoPP
    │   │   └── Other process → Collect 'show tech-support' for TAC
    │   ├── RP normal, QFP drops high → Data plane overload
    │   │   └── Check traffic rates, ACL complexity, NAT sessions
    │   └── Both high → Platform capacity exceeded, reduce load
    │
    ├── [NX-OS] CPU issue?
    │   ├── Check VDC context → Are you in the correct VDC?
    │   ├── System resources CPU high
    │   │   ├── netstack high → Forwarding overload, check ARP/routes
    │   │   ├── bgp high → Route churn, check table size and peers
    │   │   ├── stp/vpc high → L2 instability, check topology
    │   │   └── Other → Collect 'show tech-support' for TAC
    │   └── Per-module issue → Specific linecard problem
    │       └── show module → Identify failed/degraded module
    │
    ├── Memory issue? (either platform)
    │   ├── [IOS-XE] Fragmentation high → Schedule reload during window
    │   ├── [NX-OS] MemFree < 15% → Identify top consumers
    │   └── Steady growth → Memory leak, collect allocation data, plan reload
    │
    ├── Interface errors? → Classify error type
    │   ├── CRC/input errors → Layer 1 (cable, optic, SFP)
    │   ├── Output drops → QoS or congestion
    │   └── [NX-OS] ASIC errors → Hardware issue, check module health
    │
    └── All within thresholds → Document clean health

Escalation Criteria

Escalate to senior engineer or TAC when:

• CPU sustained > 90% for 15+ minutes with no identifiable cause
• Memory below 15% free with no recent change to explain consumption
• Traceback, CPUHOG, or MALLOCFAIL in logs (IOS-XE); core dump present (NX-OS)
• Any environmental alarm (power, fan, temperature)
• More than 3 routing neighbor state changes in the last hour
• NX-OS module in non-OK state or vPC peer-link failure
• QFP drops exceeding 10,000/sec with no traffic anomaly explanation (IOS-XE)

Report Template

text

DEVICE HEALTH REPORT
====================
Device: [hostname]
Platform: [IOS-XE | NX-OS]
Model: [PID from inventory]
Software: [version]
Uptime: [uptime string]
VDC: [VDC name — NX-OS only, omit for IOS-XE]
Check Time: [timestamp]
Performed By: [operator/agent]

SUMMARY: [HEALTHY | WARNING | CRITICAL | EMERGENCY]

FINDINGS:
1. [Severity] [Component] — [Description]
   Platform: [IOS-XE | NX-OS]
   Observed: [metric value]
   Threshold: [normal/warning/critical range]
   Action: [recommended action]

2. ...

PLATFORM-SPECIFIC NOTES:
- [IOS-XE: QFP drop summary or NX-OS: VDC/module status]

RECOMMENDATIONS:
- [Prioritized action list]

NEXT CHECK: [date based on severity — CRITICAL: 24hr, WARNING: 7d, HEALTHY: 30d]

Troubleshooting

Device Unresponsive to SSH

Try console access. If console also unresponsive, check power and environment via out-of-band management (CIMC for UCS-based, smart PDU, or console server). On IOS-XE: collect dir crashinfo: after recovery. On NX-OS: show cores and show system reset-reason.

NX-OS VDC Context Confusion

Health data is VDC-scoped. If metrics look wrong, verify VDC: show vdc current. Switch VDC with switchto vdc [name]. The default VDC contains system-level resources; admin VDC shows aggregate hardware.

CPU Spikes During Health Check

Show commands and SNMP polling can briefly spike CPU. Wait 30 seconds after connecting before collecting CPU data. Use terminal length 0 (IOS-XE) or terminal length 0 (NX-OS) to avoid paging pauses.

IOS-XE QFP Drops vs RP CPU

High QFP drops with normal RP CPU means the data plane is stressed but control plane is healthy. Do not restart the device — the RP is functioning normally. Investigate traffic patterns, ACL complexity, and NAT session counts. Use show platform hardware qfp active feature to identify which QFP feature is consuming resources.

NX-OS Module Issues

If show module reports a module not in "ok" state, collect show module internal exception-log module [n] before taking action. Powercycling a linecard (poweroff module [n] / no poweroff module [n]) can resolve transient failures but requires change control.

Inconsistent Memory Between Platforms

IOS-XE and NX-OS report memory differently. IOS-XE show memory statistics shows IOS memory pools; NX-OS show system resources shows Linux-level memory. NX-OS baseline memory usage is higher (55–65% is normal) because the Linux kernel and system daemons consume more than IOS-XE's monolithic process model.