Skip to main content

Performance problems

Use this information to solve performance problems.

Follow the suggested actions in the order in which they are listed in the Action column until the problem is solved.

The performance problems table is a two-column table that lists the symptoms and corrective actions for performance problems. Column 1 lists the symptom. Column 2 describes the corrective actions for each symptom.

  • Follow the suggested actions in the order in which they are listed in the Action column until the problem is solved.
  • See Parts listing, Types 7167 and 2590 to determine which components are customer replaceable units (CRU) and which components are field replaceable units (FRU) that must be replaced only by a trained service technician.
  • If an action step is preceded by (Trained service technician only), that step must be performed only by a trained service technician.
SymptomAction
Poor or slow network performance
  1. Isolate which network is operating slowly (such as storage, data, or management). You might find it helpful to use ping tools or operating-system tools such as task manager or resource manager.
  2. Check for traffic congestion on the network.
  3. Update the NIC device driver or the storage device controller device driver.
  4. Use the traffic-diagnostic tools that are provided by the I/O module manufacturer.
Operating-system performance
  1. If you have recently made changes to the compute node (for example, updated device drivers or installed software applications) remove the changes.
  2. Check for any networking issues.
  3. Check the operating-system logs for performance-related errors.
  4. Check for events related to high temperatures and power issues, because the compute node might be throttled to help with cooling. If it is throttled, reduce the workload on the compute node to help improve performance.
  5. Check for events related to disabled DIMMs. If you do not have enough memory for the application workload, your operating system will have poor performance.
  6. Ensure that the workload is not too high for the configuration.