On 8 January 2013 07:06, Preeti U Murthy preeti@linux.vnet.ibm.com wrote:
On 01/07/2013 09:18 PM, Vincent Guittot wrote:
On 2 January 2013 05:22, Preeti U Murthy preeti@linux.vnet.ibm.com wrote:
Hi everyone, I have been looking at how different workloads react when the per entity load tracking metric is integrated into the load balancer and what are the possible reasons for it.
I had posted the integration patch earlier: https://lkml.org/lkml/2012/11/15/391
Essentially what I am doing is: 1.I have disabled CONFIG_FAIR_GROUP_SCHED to make the analysis simple 2.I have replaced cfs_rq->load.weight in weighted_cpuload() with cfs.runnable_load_avg,the active load tracking metric. 3.I have replaced se.load.weight in task_h_load() with se.load.avg.contrib,the per entity load tracking metric. 4.The load balancer will end up using these metrics.
After conducting experiments on several workloads I found out that the performance of the workloads with the above integration would neither improve nor deteriorate.And this observation was consistent.
Ideally the performance should have improved considering,that the metric does better tracking of load.
Let me explain with a simple example as to why we should see a performance improvement ideally:Consider 2 80% tasks and 1 40% task.
With integration:
40%80% 40% cpu1 cpu2
The above will be the scenario when the tasks fork initially.And this is a perfectly balanced system,hence no more load balancing.And proper distribution of loads on the cpu.
Without integration
40% 40% 80% 40% 80% 40% cpu1 cpu2 OR cpu1 cpu2
Because the view is that all the tasks as having the same load.The load balancer could ping pong tasks between these two situations.
When I performed this experiment,I did not see an improvement in the performance though in the former case.On further observation I found that the following was actually happening.
With integration
Initially 40% task sleeps 40% task wakes up and select_idle_sibling() decides to wake it up on cpu1
40% -> -> 40%80% 40% 80% 40% 80% 40% cpu1 cpu2 cpu1 cpu2 cpu1 cpu2
This makes load balance trigger movement of 40% from cpu1 back to cpu2.Hence the stability that the load balancer was trying to achieve is gone.Hence the culprit boils down to select_idle_sibling.How is it the culprit and how is it hindering performance of the workloads?
*What is the way ahead with the per entity load tracking metric in the load balancer then?*
In replies to a post by Paul in https://lkml.org/lkml/2012/12/6/105, he mentions the following:
"It is my intuition that the greatest carnage here is actually caused by wake-up load-balancing getting in the way of periodic in establishing a steady state. I suspect more mileage would result from reducing the interference wake-up load-balancing has with steady state."
"The whole point of using blocked load is so that you can converge on a steady state where you don't NEED to move tasks. What disrupts this is we naturally prefer idle cpus on wake-up balance to reduce wake-up latency. I think the better answer is making these two processes load balancing() and select_idle_sibling() more co-operative."
I had not realised how this would happen until I saw it happening in the above experiment.
Based on what Paul explained above let us use the runnable load + the blocked load for calculating the load on a cfs runqueue rather than just the runnable load(which is what i am doing now) and see its consequence.
Initially: 40% task sleeps
40%80% 40% -> 80% 40% cpu1 cpu2 cpu1 cpu2
So initially the load on cpu1 is say 80 and on cpu2 also it is 80.Balanced.Now when 40% task sleeps,the total load on cpu2=runnable load+blocked load.which is still 80.
As a consequence,firstly,during periodic load balancing the load is not moved from cpu1 to cpu2 when the 40% task sleeps.(It sees the load on cpu2 as 80 and not as 40). Hence the above scenario remains the same.On wake up,what happens?
Here comes the point of making both load balancing and wake up balance(select_idle_sibling) co operative. How about we always schedule the woken up task on the prev_cpu? This seems more sensible considering load balancing considers blocked load as being a part of the load of cpu2.
Hi Preeti,
I'm not sure that we want such steady state at cores level because we take advantage of migrating wake up tasks between cores that share their cache as Matthew demonstrated. But I agree that reaching such steady state at cluster and CPU level is interesting.
IMHO, you're right that taking the blocked load into consideration should minimize tasks migration between cluster but it should no prevent fast task migration between cores that share their cache
True Vincent.But I think the one disadvantage even at cpu or cluster level is that when we consider blocked load, we might prevent any more tasks from being scheduled on that cpu during periodic load balance if the blocked load is too much.This is very poor cpu utilization
The blocked load of a cluster will be high if the blocked tasks have run recently. The contribution of a blocked task will be divided by 2 each 32ms, so it means that a high blocked load will be made of recent running tasks and the long sleeping tasks will not influence the load balancing. The load balance period is between 1 tick (10ms for idle load balance on ARM) and up to 256 ms (for busy load balance) so a high blocked load should imply some tasks that have run recently otherwise your blocked load will be small and will not have a large influence on your load balance
Also we can consider steady states if the waking tasks have a specific waking pattern.I am not sure if we can risk hoping that the blocked task would wake up soon or would wake up at time 'x' and utilize that cpu.
Ok, so you don't consider to use blocked load in load balancing any more ?
regards, Vincent
Vincent
Regards Preeti U Murthy