[Komal Thareja]

Correction: Both the CERN and CIEN racks have ConnectX-6 and GPU available on the same host. However, the CIEN rack is currently under maintenance as it is being transported back from SC.

You can proceed with your experiment on the CERN rack, subject to its availability.

Additionally, here’s a Fablib code snippet to help you check for specific resources on hosts:


fields=['name','nic_connectx_6_capacity','nic_connectx_5_capacity','tesla_t4_capacity','rtx6000_capacity', 'a30_capacity', 'a40_capacity']
output_table = fablib.list_hosts(fields=fields)


Thanks,
Komal

1 user thanked author for this post.

Tanay Maheshwari

[Komal Thareja]

Hi Tanay,

As you mentioned, the current infrastructure supports GPUs with ConnectX-5. Unfortunately, GPUs with ConnectX-6 are not a feasible option at this time. I hope the available setup works well for your experiment.

Thanks,
Komal

[Komal Thareja]

Shared the BitFile flash status with Nishanth over email. He confirmed that FPGA slice with EsNet Sites are working as expected.

Thanks,
Komal

[Komal Thareja]

Hi Vaneshi,

We are working to update the Permissions, but in the current release, you need permission Net.PortMirroring for InSlice PortMirroring to work.

Thanks,
Komal

[Komal Thareja]

Hi Vaneshi,

Your project would need Net.PortMirroring permission for this to work. Could you please check if your project has this permission? If not, Please request your Project Owner or Lead to request for these permissions from the portal.
More details for requesting the permissions can be found here.

Thanks,
Komal

[Komal Thareja]

Hi Nishanth,

I will check other sites too, it would be helpful if you can share sites you tried.
Could you please delete your slice and try again?

It might be related to which bitfile has been used to Flash the FPGA. For instance, INDI is flashed with a bit file compatible with NEU workflow.

Thanks,
Komal

• This reply was modified 1 month, 3 weeks ago by Komal Thareja.

[Komal Thareja]

Hi Yuanjun,

Details about the VM profiles are available here. For specific flavor information, please refer to the GitHub link. Feel free to reach out if you have any questions or concerns.

Thanks,

Komal

[Komal Thareja]

Hi Sourya,

I can confirm both node: MICH_3D and network interconnect4 have been deleted and are in Closed State.
In order to move the slice from ModifyError to StableError state, please execute the following block of code:


slice = fablib.get_slice(name=”Tailscale_Mesh_VPN”)
slice.modify_accept()


This should allow you to do any additional modifications.

Thanks,
Komal

[Komal Thareja]

Hi Ilya,

I’ve confirmed that the FPGA is currently allocated to a slice belonging to another user. While the FPGA may have been flashed with your bitfile, our current allocation system does not reserve FPGAs for projects based on bitfiles flash requests. If the FPGAs are not linked to a specific slice, they remain available for other users to request and utilize.

We are actively working on enhancing our allocation system and quota management, and I will take your experience as valuable feedback for these improvements. Apologies for any inconvenience this may have caused.

Thanks,
Komal

[Komal Thareja]

KANS and LOSA both have the FPGA allocated.
WASH seems to have the FPGA available but based on the Core/Ram/Disk requested, the slice might have been rejected.
Snapshot for WASH:

[Komal Thareja]

Worked with Hemil over a zoom meeting and was able to resolve the issue by renaming the bastion key in fabric_rc and re-executing the configure_and_validate.ipynb

Thanks,
Komal

[Komal Thareja]

Hi Hemil,

Could you please run the jupyter-examples-*/configure_and_validate/configure_and_validate.ipynb notebook?
This shall resolve any SSH key issues by renewing the expired bastion keys if any.

Try your setup.sh script after that and let us know if you still see this error.
In addition, could you please try to SSH to the VMs using the command shown in SSH Command coloumn.

Regarding auto configuring the IP addresses, please specify the subnet when creating a network and set mode to auto for the interfaces at slice creation. Please refer to one of the following examples for more details.

https://github.com/fabric-testbed/jupyter-examples/blob/main/fabric_examples/fablib_api/create_l2network_basic/create_l2network_basic_auto.ipynb

https://github.com/fabric-testbed/jupyter-examples/blob/main/fabric_examples/fablib_api/create_l2network_wide_area/create_l2network_wide_area_auto.ipynb

https://github.com/fabric-testbed/jupyter-examples/blob/main/fabric_examples/fablib_api/create_l3network_fabnet_ipv4/create_l3network_fabnet_ipv4_auto.ipynb

Please let us know if you still run into errors or questions.

Thanks,
Komal

[Komal Thareja]

Hi Ilya,

Could you please try your slice again? There were leaked slivers. I have cleared them, slice provisioning should work now.

Thanks,
Komal

[Komal Thareja]

Hi Ali,

We only provide 1G storage to users on Jupyter Containers in the /home/fabric/work directory. Could you please clarify or add screenshot of the df -h from the terminal in your container and also share how you are trying to upload the files?

I tried and uploaded 800 MB file to my container using the Jupyter Hub upload interface without issues.

Thanks,
Komal

[Komal Thareja]

Slice has been deleted.

Thanks,
Komal

1 user thanked author for this post.

Hyun Suk Bang

[Author]

Posts