Dear Ildiko Vancsa 

 

Thank you for your reply.

First of all, we have already deployed our pod to the same server using WRCP. (with isolcpus & intel_acc100_fec)

 

The server is trying to query the same pod with Starlingx 5.0 while supporting the resources.

 

After 5.0 installation according to StarlingX Doc, cpu is set to Application-isolate in the same way, but the resources are not visible.

 

'intel.com/intel_acc100_fec: ‘16’’
'windriver.com/isolcpus: 24’
Is there anything to do separately from starlingsX to make the above two resources available?

 

for Starlingx Installation, I did just below,

https://docs.starlingx.io/deploy_install_guides/r5_release/bare_metal/aio_simplex_install_kubernetes.html

 

for resources, I did belows,

https://protect2.fireeye.com/v1/url?k=67ad2fd7-383616cf-67aca498-0cc47a314e9a-5e0aef83dc32c5cb&q=1&e=5ed23c51-ae04-476f-a95d-f2d4d2a9c1cc&u=https%3A%2F%2Fdocs.starlingx.io%2Fnode_management%2Fkubernetes%2Fhardware_acceleration_devices%2Fset-up-pods-to-use-sriov.html
https://protect2.fireeye.com/v1/url?k=33fc70ff-6c6749e7-33fdfbb0-0cc47a314e9a-bc617f111d97e7de&q=1&e=5ed23c51-ae04-476f-a95d-f2d4d2a9c1cc&u=https%3A%2F%2Fdocs.starlingx.io%2Fr%2Fstx.5.0%2Fusertasks%2Fkubernetes%2Fuse-application-isolated-cores.html

 

When I check node description, there is no isolcpus and intel_acc100_fec.

Capacity:
  cpu:                           64
  ephemeral-storage:             10190100Ki
  hugepages-1Gi:                 32Gi
  hugepages-2Mi:                 0
  intel.com/pci_sriov_net_ca:    7
  intel.com/pci_sriov_net_f1c:   4
  intel.com/pci_sriov_net_f1u:   4
  intel.com/pci_sriov_net_fh0:   4
  intel.com/pci_sriov_net_fh0m:  4
  intel.com/pci_sriov_net_fh1:   3
  intel.com/pci_sriov_net_fh1m:  4
  memory:                        131264360Ki
  pods:                          110

  Same Server with WRCP case, we can see like this

Capacity:
  cpu:                           64
  ephemeral-storage:             10190100Ki
  hugepages-1Gi:                 32Gi
  hugepages-2Mi:                 0
  intel.com/intel_acc100_fec:    1
  intel.com/pci_sriov_net_ca:    7
  intel.com/pci_sriov_net_f1c:   4
  intel.com/pci_sriov_net_f1u:   4
  intel.com/pci_sriov_net_fh0:   4
  intel.com/pci_sriov_net_fh0m:  4
  intel.com/pci_sriov_net_fh1:   3
  intel.com/pci_sriov_net_fh1m:  4
  memory:                        131365668Ki
  pods:                          110
  windriver.com/isolcpus:        54

Please let me know if there is anything I can do.

 

Thanks and Best Regards,
Taejun

 

--------- Original Message ---------

Sender : Ildiko Vancsa <ildiko.vancsa@gmail.com>

Date : 2021-10-19 03:56 (GMT+9)

Title : Re: [Starlingx-discuss] How to Set 'isolcpus & HW Accelarator' on Starlingx 5.0

 

Dear Kim Tae-jun,

Thank you for reaching out with your question.

I think you may need to modify the yaml file to adjust it to the resources available in your system. Which means to modify these entries:
'intel.com/intel_acc100_fec: ‘16’’
'windriver.com/isolcpus: 24’

Do you have the required resources available in your environment to use this feature?

Thanks and Best Regards,
Ildikó


> On Oct 12, 2021, at 18:02, 김태준 <tj85.kim@samsung.com> wrote:
> 
> Dear starlingx members
>  
> Hello. This is Kim Tae-jun from Samsung Electronics.
> We are trying a trial to apply Starlingx to our system.
> 
> Among them, we are trying to set isolcpus and hw indicator according to the manual in starlingx docs and modify pod.
> However, even if you create yaml as manual, you can not run properly because you do not have the following resources.
>  
>   Type     Reason            Age               From               Message
>   ----     ------            ----              ----               -------
>   Warning  FailedScheduling  6s (x5 over 15s)  default-scheduler  0/1 nodes are available: 1 Insufficient intel.com/intel_acc100_fec, 1 Insufficient windriver.com/isolcpus.
>  
> resources corresponding to the capacity of the node description are not visible.
>  
> I set it by referring to the links below. I would appreciate it if you would let me know if there is anything missing in the preliminary work or what to refer to.
>  
> https://protect2.fireeye.com/v1/url?k=67ad2fd7-383616cf-67aca498-0cc47a314e9a-5e0aef83dc32c5cb&q=1&e=5ed23c51-ae04-476f-a95d-f2d4d2a9c1cc&u=https%3A%2F%2Fdocs.starlingx.io%2Fnode_management%2Fkubernetes%2Fhardware_acceleration_devices%2Fset-up-pods-to-use-sriov.html
> https://protect2.fireeye.com/v1/url?k=33fc70ff-6c6749e7-33fdfbb0-0cc47a314e9a-bc617f111d97e7de&q=1&e=5ed23c51-ae04-476f-a95d-f2d4d2a9c1cc&u=https%3A%2F%2Fdocs.starlingx.io%2Fr%2Fstx.5.0%2Fusertasks%2Fkubernetes%2Fuse-application-isolated-cores.html
>  
> Thanks you.
>  
> <Mail Attachment.gif> 
> _______________________________________________
> Starlingx-discuss mailing list
> Starlingx-discuss@lists.starlingx.io
> https://protect2.fireeye.com/v1/url?k=dd734e99-82e87781-dd72c5d6-0cc47a314e9a-0b594a17d9dfec9d&q=1&e=5ed23c51-ae04-476f-a95d-f2d4d2a9c1cc&u=http%3A%2F%2Flists.starlingx.io%2Fcgi-bin%2Fmailman%2Flistinfo%2Fstarlingx-discuss