A VCS SRDF resource OnlineTimeout limit is reached during a failback to the R1, when the R1 devices are in an updated state

book

Article ID: 100033230

calendar_today

Updated On:

Description

Steps to Reproduce

  1. Switch the service group from R1 to R2. The R1 devices are now in a "Failed over" state.
  2. Run symrdf update. The R1 devices are now in an "Updated" state.
  3. Attempt to switch the service group from R2 to R1. The SRDF agent online timeout threshold is reached because the symrfg -failedover verify command never completes.

Error Message

VCS ERROR V-16-2-13065 ([hostname]) Agent is calling clean for resource([resource_name]) because online did not complete within the expected time

Workaround

Modify /opt/VRTSvcs/bin/SRDF/SRDFAgent.pm as shown below:

Original Script:
my $ret = run_cmd("$symrdf $self->{'type'} $self->{'name'} $self->{'rdfgopt'} -c $halfCount -i 5 -failedover verify > $DEVNULL 2>&1", 0, 1, $func);
   
Modify $halfCount to 2, as shown below:
my $ret = run_cmd("$symrdf $self->{'type'} $self->{'name'} $self->{'rdfgopt'} -c 2 -i 5 -failedover verify > $DEVNULL 2>&1", 0, 1, $func);

Resolution

VCS Agent_Pack_3Q2016 contains a code change that adds an alternate method of checking the failed over status.

We recommend checking whether symrdf verify completes successfully: 
 # /opt/emc/SYMCLI/bin/symrdf  -g  [EMC SRDF DG Name]  -c 10 -i 5 -failedover verify 
 
Note how long the command takes to complete. If a verify takes more than 5 seconds, check with EMC to determine why it takes so long. 
 

Issue/Introduction

A VCS (Veritas Cluster Server) SRDF resource OnlineTimeout is reached during a failback to the R1, when the R1 devices are in an updated state.
 

Steps to Reproduce

  1. Switch the service group from R1 to R2. The R1 devices are now in a "Failed over" state.
  2. Run symrdf update. The R1 devices are now in an "Updated" state.
  3. Attempt to switch the service group from R2 to R1. The SRDF agent online timeout threshold is reached because the symrfg -failedover verify command never completes.