[ClusterLabs] Migration-threshold, timeout and interval

[Ken Gaillot]

On 10/19/2016 06:06 AM, Marcos Renato da Silva Junior wrote:
> Hi,
> 
> 
> It would be correct to say that both options have the same function :
> 
> 
> meta migration-threshold="3" op monitor interval=10s
> 
> op monitor interval=10 timeout=30
> 
> 
> If not what is the difference?

You have two separate configuration items here, the migration-threshold
and the op monitor.

The migration-threshold is the number of times the cluster will try
restarting a failed resource on any one node before marking that node
ineligible to serve the resource. The default is INFINITY (which is
actually interpreted as 1,000,000).

The monitor operation is the recurring check of the resource's status.
interval= 10 or 10s is equivalent, and means the check will be done
every 10 seconds. timeout=30 means the check will be killed (and failed)
if it takes longer than 30 seconds. IIRC the default timeout is 20 seconds.

> 
> 
> My resource config :
> 
> crm configure primitive slapd_mirrormode ocf:heartbeat:slapd params\
>  slapd="/usr/sbin/slapd" \
>  config="/etc/ldap/slapd.d/" \
>  user="openldap" group="openldap" \
>  services="ldap:///" \
>  watch_suffix="dc=mrsj,dc=lab" \
>  meta migration-threshold="3" \
>  op monitor interval=10s
> 
> 
> 
> Thanks,
>