PowerPatth path_latancy_threshold behavior

Question

I see that setting path_latencty_monitor and path_latency_threshold log events for the path latency surpassing threshold. How does this effect failover while in BasicFailover policy? Does failover happen on the first or Nth threshold failure or is this simply for monitoring purposes?

I am trying to tune Redhat cluster suite per Redhat recomentations found at:
http://kbase.redhat.com/faq/FAQ_46_13315.shtm

These examples apply to multipathd instead of powerpath and I am trying to translate this to my enviroment. multipathd lets you define a polling interval and number of allowd retries. So the failover time is polling interval * retries. I'm having difficulty finding documentation about what parameters control PowerPath failover time.

My Environment:
Redhat Enterprise Linux: 5.2, x86 / Redhat Cluster Suite 5
PowerPath for Linux 5.1.2
Storage: Clariion CX3-80

Thanks for any input!
- Charles

Conor · Answer

Hi Charles,1. path_latencty_monitor and path_latency_threshold are simply audit logs, extra logging that PowerPath provides to assist with identifying performance issues on a host, they do not influence failover in any way.2. The trigger that causes PowerPath to failover is usually the HBA timeout value, for example, with Emulex HBA's the value lpfc_nodev_tmo=10 Specifies the amount of time [0-255] that all I/O errors will be held by the HBA driver on devices that disappear. For PowerPath we recommend a value of 10 seconds. After 10 seconds the HBA driver reports to PP that the device has not responded, and PowerPath will then take action and re-route the IO to the alternate path. QLogic have similar parameters, please see the QLogic and Emulex documentation for your HBA/Driver combination.

Conor · Answer

Let us know if this has answered your question bu marking it 'Answered' thanksConor

PowerPath

PowerPatth path_latancy_threshold behavior

Was this post helpful?