Network Unreachable

[Henry XU]

Hello everyone,

I'm currently facing the network unreachable problem. There is usually no such problem, but recently when I try to ssh to the machine, I sometimes cannot make the connection. This happened yesterday and is happening again now. Is something going on at the cloudlab?

Thanks,

Henry Xu

[Mike Hibler]

There is no scheduled disruption going on. Which machine? Can you reach the
portal or other machines?

> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email
> to cloudlab-user...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/
> cloudlab-users/4209b238-4255-497d-8019-12b531b4acb4%40googlegroups.com.

[Henry XU]

Hi Mike,

I'm trying to work on c240g5-pytorch at the Wisconsin cluster. I started an experiment and was able to ssh to it yesterday, but today I cannot reach it. Strangely, I also tried to ssh to c240g5-pytorch via other people's account, and it worked. 

Is there anything wrong with my machine or my account?

Thanks,

Henry Xu

On Tuesday, November 19, 2019 at 10:03:01 AM UTC-6, Mike Hibler wrote:

There is no scheduled disruption going on. Which machine? Can you reach the
portal or other machines?

On Tue, Nov 19, 2019 at 07:55:12AM -0800, 'Henry XU' via cloudlab-users wrote:
> Hello everyone,
>
> I'm currently facing the network unreachable problem. There is usually no such
> problem, but recently when I try to ssh to the machine, I sometimes cannot make
> the connection. This happened yesterday and is happening again now. Is
> something going on at the cloudlab?
>
> Thanks,
> Henry Xu
>
> --
> You received this message because you are subscribed to the Google Groups
> "cloudlab-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email

> to cloudla...@googlegroups.com.

[Mike Hibler]

The short answer is that the Internet-facing control network interface
was not configured. We do custom configuration of that interface at boot
time, so if anything messes with network setup in the boot path, this
failure can happen.

Specifically, on Ubuntu 16 we do not use the NetworkManager and it appears
that you installed it either directly or indirectly and that caused the
problem. I "fixed" this by disabling the NetworkManager:

systemctl disable NetworkManager

so the current node will continue to work. However, you will have to update
your custom image if you want future experiments to work correctly.

Note that you can always login to the node via its console if this happens
again.

> to cloudlab-user...@googlegroups.com.

> To view this discussion on the web visit https://groups.google.com/d/msgid/

> cloudlab-users/0f5247d9-816f-4558-a071-6d1f6478e82c%40googlegroups.com.

[Henry XU]

Thank you very much!