[Mert Cevik]

Maintenance on FABRIC-MASS is completed.

[Mert Cevik]

This outage is resolved. All FABRIC services are operational and new requests can be submitted.

[Mert Cevik]

This outage is resolved. VMs on FABRIC-NCSA rack are reachable from public internet. Please let us know if you encounter any problems.

[Mert Cevik]

We are working with the hosting campus for a known network issue that may require part replacement. No ETA is available, updates will be posted when the problem is resolved.

[Mert Cevik]

If you create an example slice to reveal the problem, we can work on it to understand better, suggest solutions, consider changes on public ACLs. Please let us know with the name of the slice and FABRIC site, if this sounds good for you.

[Mert Cevik]

Does it mean that TACC site is not support IPv6?

Yes, TACC site does not support IPv6, it supports IPv4 addresses.

Sites MAX, TACC (and upcoming MASS) support only IPv4 addresses.

Sites STAR, MICH, UTAH, NCSA, WASH, DALL, SALT support only IPv6 addresses.

• This reply was modified 2 years, 11 months ago by Mert Cevik.

[Mert Cevik]

Hi Adam,

I checked FABRIC-MAX, however I could not repeat the problem. It will make it easier for us if you can share information about the slice (eg. topology information as a list/description, quick drawing,  request python code etc).

Mert

[Mert Cevik]

Network outage is resolved. FABRIC-TACC is online.

[Mert Cevik]

This maintenance is re-scheduled for Tuesday 1/11/22 –  9am-12pm ET.

[Mert Cevik]

I cannot comment on that without diving into the logs to see how the VM was created, PCIe device attached etc. Instead, I suggest starting a new slices (including the fixes for management network), then step by step checking the status with respect to the requested devices. I will be able to help if you prefer this approach.

[Mert Cevik]

I checked one of your VMs on FABRIC-MAX.

Name: ff5acfa1-bbff-44a0-bf28-3d7d2f038d1f-Node1
IP: 63.239.135.79

In your workflow to configure slice, you change the network settings that affect Management Network.

[root@node1 ~]# systemctl status NetworkManager
● NetworkManager.service – Network Manager
Loaded: loaded (/usr/lib/systemd/system/NetworkManager.service; disabled; ve>
Active: inactive (dead)
Docs: man:NetworkManager(8)
[root@node1 ~]# systemctl is-enabled NetworkManager
disabled

Interface eth0 should persist its IP address configuration (from RFC1918 subnet). Network node of the virtualization platform control external traffic either by NAT’ing or routing against the configured IP address. Currently you have the following:

[root@node1 ~]# ifconfig -a
docker0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500
inet 172.17.0.1 netmask 255.255.0.0 broadcast 172.17.255.255
ether 02:42:86:f0:f7:a8 txqueuelen 0 (Ethernet)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

eth0: flags=4098<BROADCAST,MULTICAST> mtu 9000
ether fa:16:3e:49:8e:5a txqueuelen 1000 (Ethernet)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING> mtu 65536
inet 127.0.0.1 netmask 255.0.0.0
inet6 ::1 prefixlen 128 scopeid 0x10
loop txqueuelen 1000 (Local Loopback)
RX packets 16 bytes 916 (916.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 16 bytes 916 (916.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

When network settings is reverted to the original for Management Network, your VM shows the following:

[root@node1 ~]# ifconfig -a
docker0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500
inet 172.17.0.1 netmask 255.255.0.0 broadcast 172.17.255.255
ether 02:42:69:1f:14:22 txqueuelen 0 (Ethernet)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

eth0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 9000
inet 10.20.4.94 netmask 255.255.255.0 broadcast 10.20.4.255
inet6 fe80::f816:3eff:fe49:8e5a prefixlen 64 scopeid 0x20 ether fa:16:3e:49:8e:5a txqueuelen 1000 (Ethernet)
RX packets 2015 bytes 232936 (227.4 KiB)
RX errors 0 dropped 31 overruns 0 frame 0
TX packets 1978 bytes 226617 (221.3 KiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING> mtu 65536
inet 127.0.0.1 netmask 255.0.0.0
inet6 ::1 prefixlen 128 scopeid 0x10
loop txqueuelen 1000 (Local Loopback)
RX packets 1160 bytes 58116 (56.7 KiB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 1160 bytes 58116 (56.7 KiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0

and it’s reachable back again.

$ ping 63.239.135.79 -c 3
PING 63.239.135.79 (63.239.135.79): 56 data bytes
64 bytes from 63.239.135.79: icmp_seq=0 ttl=52 time=23.257 ms
64 bytes from 63.239.135.79: icmp_seq=1 ttl=52 time=21.347 ms
64 bytes from 63.239.135.79: icmp_seq=2 ttl=52 time=17.025 ms

— 63.239.135.79 ping statistics —
3 packets transmitted, 3 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 17.025/20.543/23.257/2.607 ms

You need to review standard installation procedures of platforms such as Docker, Kubernetes, OpenStack and consider changes for the Management Network of your slivers.

[Mert Cevik]

Hello Tejasri,

Can you try to login to the JupyterHub again and let us know?

I’m not the best person to help with this at the moment, but I will be able to contact the team. It will be useful to know how multiple login trials work.

Mert

[Author]

Posts