A possible timing issue:when using MultiNICB with mpathd mode.

book

Article ID: 100017041

calendar_today

Updated On:

Description

Error Message

N/A

 

Cause

With mpathd enabled, IPMultiNICB sometime detects IP faults, while mpathd is just doing some changes beneath, such as IP failover/failback.  MultiNICB does not detect any failure for such IP operations by mpathd, but MultiNICB exports this information to IPMultiNICB. A small possibility exists that, while mpathd is applying changes, MultiNICB exports that information during its monitor cycle and the IPMultiNICB monitor may detect IP faults if the IP is down in this exported information. The MonitorInterval of MultiNICB is 10 seconds as default, therefore, the next MultiNICB monitor will see the right IP status, and this will be enforced by the next monitor of IPMultiNICB.


For example, when pulling one of the cables to check the behavior of IPMultiNICB/MultiNICB with mpathd, it will sometimes report unexpected offline from the IPMultiNIC resource.

Resolution

To avoid such a timing issue, set the resource attributes OnlineRetryLimit=1 and ToleranceLimit=1 for IPMultiNICB when using MultiNICB in mpathd mode.

 

Issue/Introduction

A possible timing issue when using MultiNICB with mpathd mode.