atolab/fog05

the infrastructure for Fog and MEC computing

marzmesas

@marzmesas

Hi @gabrik , another quick question. At this link https://fog05.io/docs/getting-started/etsi-osm-interop/ , you guys establish one of the vim account parameters as <zenoh server ip>:7447 , yet at this link from the OSM site https://osm.etsi.org/wikipub/index.php/Eclipse_fog05#Upload_Images we can find the same parameter as <rest proxy ip>:8080
Could you clarify for me which of these two configurations would be correct? Thank you!

Gabriele Baldoni

@gabrik

Hi @marzmesas, you can use the one with <zenoh server ip>:7447 regarding the api you should only pass the image descriptor, so it will be api.image.add(img_descriptor)

marzmesas

@marzmesas

Hi @gabrik , is the zenoh server ip the same as the yaks ip? If not, where can I find such ip address?

Gabriele Baldoni

@gabrik

Yes it is the same

marzmesas

@marzmesas

Perfect, I had another question regarding the VNF. The only thing to be packaged into a tar.gz in order to instantiate the VNF would be the alpine-vnf.yaml , where the name of the image would be the name of the image we have previously uploaded? That's the way?

Gabriele Baldoni

@gabrik

Yes that’s the way

marzmesas

@marzmesas

Regarding this, in the OSM Fog05 link ( https://osm.etsi.org/wikipub/index.php/Eclipse_fog05#Upload_Images ) we can find that the image is referenced under a different name as the one in the fog05 website. In the OSM website I think they reference it with the uuid of the image. Is that the way to reference the image in the vnfd yaml?

Gabriele Baldoni

@gabrik

It should work in both ways, but using the name is the correct way in general

marzmesas

@marzmesas

Hi @gabrik , I was trying to upload the image but I came around a dumb question. In order to upload the image, can we do it through the bash command line or do we have to run a python script to access the FDU api?

Gabriele Baldoni

@gabrik

Hi @marzmesas at this time there is no command line interface so you need to use the APIs.

marzmesas

@marzmesas

So we pretty much should use a script like the one created to run the helloworld example?

Gabriele Baldoni

@gabrik

Yes, but you just need to register the image, using the api.image.add API

marzmesas

@marzmesas

yes, so it'd basically be a python program that ran that instruction

Gabriele Baldoni

@gabrik

Yes, excatly

marzmesas

@marzmesas

Where can I find the methods of the api.image class ? so I can know how to get a log printed out of the result of api.image.add

Gabriele Baldoni

@gabrik

API docs are available on readthedocs.io -> https://eclipse-fog05-python3-client-api.readthedocs.io/en/latest/fimapi.html#image

marzmesas

@marzmesas

Hi Gabriele, after trying some things I got this error coming up. Here are my descriptor and the python script to upload the image as well as the error popping up when executing the script

Do you have any ideas on how to solve this?

Gabriele Baldoni

@gabrik

You have to pass the dict object to the API not the descriptor file, so you should read the file content into a dict and then pass it to the API

Like it is done for the HelloWorld example

Gabriele Baldoni

@gabrik

So it should be like this:

import json
from fog05 import FIMAPI 

def read_file(filepath):
    with open(filepath, 'r') as f:
        data = f.read()
    return data

api = FIMAPI()
Img_desc = json.loads(read_file(<path to json descriptor>))
res = api.image.add(img_desc)
print(res)
api.close()

marzmesas

@marzmesas

Hi @gabrik , this is the script I ran, with the exact descriptor y previously sent

And these were the errors it reported

Could you help me out with that?

Gabriele Baldoni

@gabrik

Can you print the content of img_desc?

marzmesas

@marzmesas

Hi @gabrik , this is what I get when I change the script for it to print img_desc

Gabriele Baldoni

@gabrik

Ok, it seems ok, because the id is added by the API if it is not present. Can you confirm you are using version 0.2.1 of the APIs?

marzmesas

@marzmesas

How can I do so?

Gabriele Baldoni

@gabrik

pip3 list | grep fog

marzmesas

@marzmesas

I checked and it seems to be running 0.2.0

Gabriele Baldoni

@gabrik

Ok, let me check for that version, maybe there was a bug

marzmesas

@marzmesas

Sure, if not just let me know how and I'll update it to 0.2.1

Gabriele Baldoni

@gabrik

Ok this was one of the issue solved with 0.2.1, to update just run

pip3 uninstall fog05==0.2.0 -y
pip3 install fog05==0.2.1

marzmesas

@marzmesas

Just installed it, ran the original program and this came up

Gabriele Baldoni

@gabrik

Strange, I’m on it

marzmesas

@marzmesas

Awesome, let me know on any updates

Thank you!

Gabriele Baldoni

@gabrik

We found a small bug in that API, the simple workaround is to populate the uuid file of the descriptor with valid uuid, you can get one using uuidgen

marzmesas

@marzmesas

Is it a class that I can find in the API?

Or could you give me an example on it by any chance?

Gabriele Baldoni

@gabrik

It is an utility in Linux, so you can just open a terminal and type uuidgen the result should be like this:

$ uuidgen
25843eb9-82c5-4bd6-b6c6-5df22766b5eb

Then copy and paste this uuid in the image descriptor in the uuid field.

It should become like this:

{ 
    “name”:”alpine”,
    “uuid”:"25843eb9-82c5-4bd6-b6c6-5df22766b5eb”,
    “uri”:”lxd://alpine/edge”,
    “checksum”:””,
    “format”:””
}

marzmesas

@marzmesas

Once I did that, the image did work

Where communities thrive

People

Repo info

Activity