IBM Power / AIX Daily Health Checks and Output Interpretation
This is a stored daily-check reference for IBM Power systems running AIX. The source runs commands as root; use the privileges required by the actual command and local policy. Its thresholds are local checklist values rather than universal IBM limits. Interpret exceptions against the exact model, workload, configuration and trend.
Environment
| Check | Source checklist value |
|---|---|
| Room temperature | 24 ± 3 °C |
| Relative humidity | 50 ± 5% |
| Supply | 200–240 V AC, single phase, within the actual equipment specification |
| Cabling | No excessive bending or loose connections |
Physical inspection
| Check | Expected condition in the source |
|---|---|
| Control-panel fault indicator | Amber exclamation indicator off |
| Drive indicators | No fault amber indication |
| NIC / Fibre Channel indicators | Appropriate link/activity indication for actual port use |
| Fans | Airflow present, no unusual noise |
System checklist
| Check | Command | Source target |
|---|---|---|
| Hardware error report | errpt -dH | more | Review hardware-class entries; the source target is no entries |
| Root-volume-group disks | lsvg -p rootvg | PV STATE active |
| Logical-volume synchronization | lsvg -l rootvg | LV state syncd (open/closed is separate) |
| Storage paths | lspath | Expected operational paths Enabled |
| Filesystem use | df -g | At most 90% |
| Paging-space use | lsps -s | At most 20% |
| Resources | topas | CPU below 85%, source memory indicator below 95% |
1. Hardware reports: errpt -dH
TIMESTAMP uses the compact month/day/hour/minute/year form MMDDhhmmYY. Type I is informational, T temporary and P permanent; P calls for investigation but is not a severity ranking independent of the full report. Class H is hardware and S software. RESOURCE_NAME identifies the reported resource; resolve its location and detailed error rather than mapping every sysplanar/system-board resource to a particular power supply or fan.
The screenshot contains type-I, class-H ARRAY CONFIGURATION CHANGED entries. An entry is not automatically a current failed part, and empty errpt output is not a complete proof of hardware health. Review DESCRIPTION, details, age and related service events.
2. Local disk state: lsvg -p rootvg
The source expects active physical volumes. Missing/removed states require checking accessibility, configuration and the actual disk/service history; the state alone does not distinguish physical failure from every other cause.
3. Logical volumes and mirrors: lsvg -l rootvg
The synchronization component should be syncd. A closed/syncd boot logical volume can be normal; closed is not the same as stale. A stale state indicates unsynchronized physical partitions and needs investigation in context. Confirm the actual mirror layout rather than assuming every logical volume is mirrored.
4. Multipath storage: lspath
Check that the expected paths are Enabled. Missing, Failed or Disabled can indicate trouble, but Disabled may be intentional. The source permits ignoring obsolete Missing paths from replaced disks only after confirming they are obsolete and that the active disk still has the required paths. Do not dismiss an unexplained Missing record.
5. Filesystem capacity: df -g
The checklist’s 90% threshold triggers capacity review, cleanup or expansion planning. Use the filesystem %Used column, not inode %Iused; pseudo-filesystems can show dashes. Evaluate actual growth and application needs before changing capacity.
6. Paging space: lsps -s
Percent Used is paging-space utilization. The source flags values above 20%, but this alone does not prove current physical-memory pressure. Examine paging activity and workload over time. The 1% screenshot and the separate topas paging-space sample are not established as simultaneous captures.
7. Resources: topas
The source target CPU utilization below 85% corresponds to Idle above about 15%. The captured Idle is 68.7%. It uses %Comp plus %Noncomp as a memory-occupancy indicator; these categories include computational and noncomputational/cache frames. A high cache-related occupancy alone does not establish a memory shortage. Read paging rates and the workload as well; do not add overlapping memory categories blindly.
The source also uses disk Busy% below 90% as a local screening value. Sustained exceptions can justify deeper performance analysis, but a single sample or threshold crossing does not prove a bottleneck. Correlate response time, queueing, errors, paging and historical baseline before concluding that hardware needs replacement or more memory.