Migration is not happening in PlanetLab Examples any specific reason?

[Riman-M]

Experiment name: 20110303_lrr_mu_1.2
Number of hosts: 800
Number of VMs: 1052
Total simulation time: 600.01 sec
Energy consumption: 5.87 kWh
Number of VM migrations: 0
SLA: NaN%
SLA perf degradation due to migration: NaN%
SLA time per active host: 0.00%
Overall SLA violation: NaN%
Average SLA violation: 0.00%
Number of host shutdowns: 800
Mean time before a host shutdown: 591.01 sec
StDev time before a host shutdown: 51.21 sec
Mean time before a VM migration: NaN sec
StDev time before a VM migration: NaN sec
Execution time - VM selection mean: 0.00000 sec
Execution time - VM selection stDev: 0.00000 sec
Execution time - host selection mean: 0.00200 sec
Execution time - host selection stDev: 0.00141 sec
Execution time - VM reallocation mean: 0.00050 sec
Execution time - VM reallocation stDev: 0.00071 sec
Execution time - total mean: 0.00600 sec
Execution time - total stDev: 0.00283 sec

[codaphobe]

Neither are they happening on the Random Workload, seems like an issue in the Power module

=============================================
Experiment name: random_iqr_mu_1.5
Number of hosts: 50
Number of VMs: 50
Total simulation time: 600.01 sec
Energy consumption: 0.38 kWh
Number of VM migrations: 0
SLA: NaN%
SLA perf degradation due to migration: NaN%
SLA time per active host: 0.00%
Overall SLA violation: NaN%
Average SLA violation: 0.00%
Number of host shutdowns: 50
Mean time before a host shutdown: 594.01 sec
StDev time before a host shutdown: 42.43 sec
Mean time before a VM migration: NaN sec
StDev time before a VM migration: NaN sec
Execution time - VM selection mean: 0.00000 sec
Execution time - VM selection stDev: 0.00000 sec
Execution time - host selection mean: 0.00000 sec
Execution time - host selection stDev: 0.00000 sec
Execution time - VM reallocation mean: 0.00050 sec
Execution time - VM reallocation stDev: 0.00071 sec
Execution time - total mean: 0.00150 sec
Execution time - total stDev: 0.00071 sec

[deRemo]

I'll have a look into it, but I believe the migration feature was not working in CloudSim 6G as well

[Riman-M]

So should we use Cloudsim 5G for that?

[codaphobe]

I've tried cloudsim 7 pre that's working fine.

[deRemo]

@codaphobe I can give it a look, unless some of you is interested in helping on that. Can you tell me which example did you try>

[Riman-M]

@codaphobe I can give it a look, unless some of you is interested in helping on that. Can you tell me which example did you try>

All examples has same issue. planetlab and random both

[codaphobe]

@codaphobe I can give it a look, unless some of you is interested in helping on that. Can you tell me which example did you try>

I am open to help but i just found a quick solution to make migrations happen by removing the following condition from the method getCompletedVms() from the HostDynamicWorkload class :

if (guest instanceof VirtualEntity vm && vm.getNumberOfGuests()==0) {
				vmsToRemove.add(vm);
			}

getCompletedVms() is called from PoweDatacenter class after all the Power data has been calculated to remove unused Vm's?

I did not understand why this condition was written since all Vms as instance of VirtualEntity and the other condition is always True unless there are nested Vm's. I could not find the commit where this part of the code was introduced to know why it was written.

[deRemo]

Hi @codaphobe, sorry for the late reply: it was added in commit cb406da to support nested Vms.. Much like the old ContainerHostDynamicWorkload class used to do for containers - that's from where I took inspiration. However, the latter class may have been bugged to begin with :p Feel free to open a PR with your findings