[Pacemaker] resources not rebalancing

[Andrew Beekhof]

On 11 Jun 2014, at 10:59 pm, Patrick Hemmer <pacemaker at feystorm.net> wrote:

> From: Andrew Beekhof <andrew at beekhof.net>
> Sent: 2014-06-11 02:36:15 EDT
> To: The Pacemaker cluster resource manager <pacemaker at oss.clusterlabs.org>
> Subject: Re: [Pacemaker] resources not rebalancing
> 
>> On 11 Jun 2014, at 3:44 pm, Patrick Hemmer <pacemaker at feystorm.net>
>>  wrote:
>> 
>> 
>>>> Right. But each node still has 4998000+ units with which to accommodate something that only requires 10000.
>>>> Thats about 0.2% of the remaining capacity, so wherever it starts, its hardly making a dint.
>>>> 
>>>> 
>>> You're thinking of the 'utilization' placement strategy.
>>> 
>> If something is "out of balance" by 0.2%... thats still quite balanced.
>> 
> 
> I agree, but that's not how the balanced strategy is documented as working. It says resources are balanced based on the amount of free capacity on each node, not based on approximate percentages.

I know, i was just saying...
You filed a bug right? We'll get it fixed, just not today.

> However, to nullify this argument, I adjusted the numbers:
>     # crm_simulate -URL
> 
>     Current cluster status:
>     Online: [ i-053f1f59 i-083f1f54 i-093f1f55 ]
> 
>      dummy1    (ocf::pacemaker:Dummy):    Started i-053f1f59 
>      Clone Set: dummy2-clone [dummy2] (unique)
>          dummy2:0    (ocf::pacemaker:Dummy):    Started i-053f1f59 
>          dummy2:1    (ocf::pacemaker:Dummy):    Started i-083f1f54 
>          dummy2:2    (ocf::pacemaker:Dummy):    Started i-093f1f55 
>          dummy2:3    (ocf::pacemaker:Dummy):    Started i-083f1f54 
>          dummy2:4    (ocf::pacemaker:Dummy):    Started i-093f1f55 
> 
>     Utilization information:
>     Original: i-053f1f59 capacity: cpu=50000 mem=3840332000
>     Original: i-083f1f54 capacity: cpu=50000 mem=3840332000
>     Original: i-093f1f55 capacity: cpu=50000 mem=3840332000
>     calculate_utilization: dummy1 utilization on i-053f1f59: cpu=40000
>     calculate_utilization: dummy2:1 utilization on i-083f1f54: cpu=2000
>     calculate_utilization: dummy2:2 utilization on i-093f1f55: cpu=2000
>     calculate_utilization: dummy2:0 utilization on i-053f1f59: cpu=2000
>     calculate_utilization: dummy2:3 utilization on i-083f1f54: cpu=2000
>     calculate_utilization: dummy2:4 utilization on i-093f1f55: cpu=2000
>     Remaining: i-053f1f59 capacity: cpu=8000 mem=3840332000
>     Remaining: i-083f1f54 capacity: cpu=46000 mem=3840332000
>     Remaining: i-093f1f55 capacity: cpu=46000 mem=3840332000
> 
> dummy1 is sharing a node with dummy2:0. The node they are on has significantly less capacity, and is almost depleted, while the others are largely underutilized.
> 
> _______________________________________________
> 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://bugs.clusterlabs.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 841 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://oss.clusterlabs.org/pipermail/pacemaker/attachments/20140612/85ffb321/attachment-0001.sig>