[Pacemaker] A question and demand to a resource placement strategy function

Andrew Beekhof andrew at beekhof.net
Tue Apr 19 09:17:21 UTC 2011


Yan is our utilization expert, lets see if he can provide some
direction here :-)

-- Andrew

2011/4/18 Yuusuke IIDA <iidayuus at intellilink.co.jp>:
> Hi, Andrew
>
> I want to disperse using a resource placement strategy function of Pacemaker-1.1
> in the fail-over point of the resource in N to N environment.
>
> After testing a function by the following setting, there was the pattern that
> the dispersion of the resource was not carried out.
> * 1ACTIVE: 2PASSIVE
> * placement-strategy=balanced
> * capacity of 2PASSIVE is a tie score
>
> * Initial state
> Online: [srv-b1 srv-b2 srv-a1]
> Full list of resources:
> main_rsc1 (ocf::pacemaker:Dummy): Started srv-a1
> main_rsc2 (ocf::pacemaker:Dummy): Started srv-a1
> main_rsc3 (ocf::pacemaker:Dummy): Started srv-a1
>
> # crm configure ptest utilization
> Utilization information:
> Original: srv-b2 capacity: capacity=3
> Original: srv-b1 capacity: capacity=3
> Original: srv-a1 capacity: capacity=3
> calculate_utilization: main_rsc1 utilization on srv-a1: capacity=1
> calculate_utilization: main_rsc2 utilization on srv-a1: capacity=1
> calculate_utilization: main_rsc3 utilization on srv-a1: capacity=1
> Remaining: srv-b2 capacity: capacity=3
> Remaining: srv-b1 capacity: capacity=3
> Remaining: srv-a1 capacity: capacity=0
>
> * When it is dispersed definitely
> When I produced trouble in a resource in order of next, I disperse and am placed
> in the node that the remainder of capacity has a big.
>
> main_rsc1 -> main_rsc2 -> main_rsc3
>
> Online: [srv-b1 srv-b2 srv-a1]
> Full list of resources:
> main_rsc1 (ocf::pacemaker:Dummy): Started srv-b1
> main_rsc2 (ocf::pacemaker:Dummy): Started srv-b2
> main_rsc3 (ocf::pacemaker:Dummy): Started srv-b1
>
> # crm configure ptest utilization
> Utilization information:
> Original: srv-b2 capacity: capacity=3
> Original: srv-b1 capacity: capacity=3
> Original: srv-a1 capacity: capacity=3
> calculate_utilization: main_rsc1 utilization on srv-b1: capacity=1
> calculate_utilization: main_rsc2 utilization on srv-b2: capacity=1
> calculate_utilization: main_rsc3 utilization on srv-b1: capacity=1
> Remaining: srv-b2 capacity: capacity=2
> Remaining: srv-b1 capacity: capacity=1
> Remaining: srv-a1 capacity: capacity=3
>
> * When it is not dispersed well
> When I produced trouble in a resource in order of next, I am partial, and the
> resource is placed in one node.
>
> main_rsc3 -> main_rsc2 -> main_rsc1
>
> Online: [srv-b1 srv-b2 srv-a1]
> Full list of resources:
> main_rsc1 (ocf::pacemaker:Dummy): Started srv-b1
> main_rsc2 (ocf::pacemaker:Dummy): Started srv-b1
> main_rsc3 (ocf::pacemaker:Dummy): Started srv-b1
>
> # crm configure ptest utilization
> Utilization information:
> Original: srv-b2 capacity: capacity=3
> Original: srv-b1 capacity: capacity=3
> Original: srv-a1 capacity: capacity=3
> calculate_utilization: main_rsc1 utilization on srv-b1: capacity=1
> calculate_utilization: main_rsc2 utilization on srv-b1: capacity=1
> calculate_utilization: main_rsc3 utilization on srv-b1: capacity=1
> Remaining: srv-b2 capacity: capacity=3
> Remaining: srv-b1 capacity: capacity=0
> Remaining: srv-a1 capacity: capacity=3
>
> I think that this problem occurs by difference in order of handling of resource.
> I attach hb_report when a problem occurred.
> Is this movement a bug?
> Or does my setting make a mistake of any?
>
> Best Regards,
> Yuusuke IIDA
> --
> ----------------------------------------
> METRO SYSTEMS CO., LTD
>
> Yuusuke Iida
> Mail: iidayuus at intellilink.co.jp
> ----------------------------------------
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker
>
>




More information about the Pacemaker mailing list