Forum Replies Created
-
Posts
-
Network services are verified to be working across the sites.
FPGA itself can be the concern. Can you share the output of
command = 'cd esnet-smartnic-fw/sn-stack/ && docker compose exec smartnic-fw sn-cli cmac status'?As an update,
according to our tests and observations since reporting the issue to the hosting site, this issue did not repeat itself and we conclude that it’s resolved.
This problem is caused by some issues within the hosting site’s and/or its upstream provider’s IPv6 routing, specific for the EDUKY node. Work is in progress to address this issue, but we don’t have a time yet for a resolution.
FABRIC provides the following VM images that have docker (and openvswitch) installed in them
– docker_ubuntu_20
– docker_ubuntu_22
– docker_ubuntu_24
– docker_rocky_8On the other hand, the issue may present itself for some other specific servers. As a workaround, the following may help for a temporary solution.
Get the IPv4 address of the server
get.docker.com –> 108.157.150.23
download.docker.com –> 108.157.150.36Adjust NAT64 addresses wrt FABRIC’s NAT64 gateway prefix
get.docker.com –> 2600:2701:5000:5001::6c9d:9617
download.docker.com –> 2600:2701:5000:5001::6c9d:9624Add to /etc/hosts
echo “2600:2701:5000:5001::6c9d:9617 get.docker.com” >> /etc/hosts
echo “2600:2701:5000:5001::6c9d:9624 download.docker.com” >> /etc/hostsThen, you should be able to reach these sites (routed to their IPv4 addresses through FABRIC’s NAT64)
ubuntu@Node1:~$ curl -I https://get.docker.com
HTTP/1.1 200 OK
Content-Type: text/plain
Content-Length: 22592
Connection: keep-alive
Date: Mon, 10 Mar 2025 08:24:07 GMT
Last-Modified: Mon, 23 Dec 2024 12:06:43 GMT
x-amz-version-id: jBhC8udY54KPazAadGzx.egHAVktsNCa
ETag: “1a91bbd5af88e59c539b2f1ac7490347”
Server: AmazonS3
X-Cache: Hit from cloudfront
Via: 1.1 3ce25b408212b0431973764f2829bcd2.cloudfront.net (CloudFront)
X-Amz-Cf-Pop: MCI50-P2
X-Amz-Cf-Id: NtXIyn1U5TQwjsaleKkIXEG7buaVTJN7vPExXgeAiqDGwVdHe9Z6pA==
Age: 17673It should be possible via fablib.get_image_names()
https://fabric-fablib.readthedocs.io/en/latest/fablib.html#fabrictestbed_extensions.fablib.fablib.fablib.get_image_namesDear Experimenters,
FABRIC-CLEM is back online. VMs are accessible, their PCI devices are re-attached, however IP addresses may need to be re-assigned by the experimenters.
We have an issue on the dataplane that is interrupting traffic for L2STS and Layer3 services. We will post updates.
Hello Yuanjun.
I don’t have the details of your slice request, but from the STAR status page on the portal, host resources for hosts that have the RTX6000 GPUs (star-w1 and star-w2) have one GPU available. Host star-w2 has a very little amount of RAM. I checked slice creation on both hosts with their available GPUs and it works well. Just, for the VM that would be placed on star-w2, I selected very small amount of RAM and CPU to fit in it.
Slice modification is supported on FABRIC. Yes you can remove the SRI VM and re-create VMs etc. It should be straightforward, but you can also check the link below for examples
MICH VM (2607:f018:110:11:f816:3eff:fe98:ba32) is online. I confirmed connection over the management network (through FABRIC bastion hosts) works well. It has an ssh key with alias “slice_key” in it (under user ubuntu)
SRI VM (192.5.67.209) is online (responding to pings), but it does not allow SSH connections that it’s supposed to allow from the controller node. SSH keys might have been modified somehow. I don’t have much to do with this VM.
