Forum Replies Created
-
Posts
-
Hello,
You can copy the ssh configuration file from work/fabric_config in Jupyter Hub to your laptop (into e.g. ~/.ssh/ directory) along with referenced bastion and sliver keys and then use SCP to copy those files onto your laptop. We are not able to provide significant storage space inside the Jupyter Hub.
This tag is not enforced so the problem is somewhere else. Please, when posting a problem begin with posting the exact error traces – we provide some guidelines and best practices for reporting problems on the forums in this article:
Assuming you are using the management network of the VM, you may find this article useful:
Reaching IPv4-only resources like GitHub or Docker Hub from IPv6 FABRIC sites
The APIs are enabled for your use, you do need to use them judiciously and keep in mind that if you can invoke them, so can someone else if they have your token.
Nishanth,
I do not believe it is possible to automate getting the token – it requires someone to login using your credentials. However the token should be good for a while – you should be able to select the lifetime.
Also please note that the token allows access to your container via API without any other checks – if someone steals it they may act on your behalf in the container so long as you have an active container (e.g. upload and execute a notebook) or wipe its contents clean. You must secure the token. Be mindful of e.g. uploading into GitHub or sharing it with others.
Make the home directory writable (or write to another directory where root can write). Note that you are doing
touchas user rocky, but capturing asroot.# make home directory world writable
[rocky@Node1 ~]$ chmod go+rwx .
# i’m using eth0 as capture interface – that doesn’t matter
[rocky@Node1 ~]$ sudo tshark -i eth0 -w logfile
Running as user “root” and group “root”. This could be dangerous.
Capturing on ‘eth0’In general we are not responsible for OS behaviors – this is left to the experimenter. Our images are standard cloud images maintained by the creators of different distributions and they exhibit standard behaviors for those distributions.
What an excellent question!
Yes JupyterHub has two levels of REST APIs – one – the JupyterHub API (for admins) the other the Notebook Server API accessible to users.
You need to obtain the API token by going to Hub Control Panel (see the image)
The APIs can be invoked using the following URL:
https://jupyter.fabric-testbed.net/user/<username>/apiwhere <username> is your primary email (in fact you should see it in your browser address field when you start the notebook, except it will look like
https://jupyter.fabric-testbed.net/user/<username>/lab)You can use cURL to make the call (this shows how to start a particular notebook, moving files in and out is similar):
curl -X POST -H "Authorization: token <token string>" "https://jupyter.fabric-testbed.net/user/<username>/api/sessions" -d '{"name":"configure_environment.ipynb","type":"notebook","path":"jupyter-examples-rel1.3.3/fabric_examples/fablib_api/configure_environment/configure_environment.ipynb"}'The owner of the project you are working in must request this permission to be added to your project via the portal.
It is quite likely that this is related to either switch behavior or the behaviour of SharedNICs. You appear to be using SharedNICs, and for most control over your experiment you should be using SmartNICs as those are not shared with anyone.
A lot of time it has to do with race conditions that exist when querying for slice state and information about e.g. IP addresses arriving asynchronously from different sites. One workaround is to call get_slice() again – that’s what queries the system for the latest slice model that has the information to produce the SSH command line (like the management IP).
get_ssh_command() just looks inside the current model and reports what is there (so while the name may suggest it gets the latest information, it doesn’t – its information is as current as the last call to get_slice()).
Let us check on this slice before you delete. We should be able to find out what happened, in some cases we are able to fix problems like this.
The notebook seems very straightforward. The problem must be in your environment. Please be sure to wait until the provisioning cell exits – however it does, whether through error, timeout or to completion. There are timeouts built into the fablib code so progress is always guaranteed.
If you are unable to fix the problem, you can sign up for office hours under ‘Experiment Help’ category.
