The interface is super slow

[Smahane]

Hello Vanessa,

I installed Tunel and I was able to lunch the GUI interface. But everything is super slow to the point that it's not usable. Any idea what I could be missing? Clicking any button takes more than 2 min to load.

Thanks

[vsoch]

What is the context where you are running it? It is likely related to that. Are you running on your host or from a forwarded display?

[Smahane]

Connected via VNC and running on the host. everything is local on a Fedora system

[vsoch]

can you take a look at how much memory your system is using / has free? One idea is that you are running out.

[Smahane]

How much memory does it need? I have plenty i believe:

[vsoch]

when you look at the container logs, is there anything iffy? Use docker ps to get the id, then to see the logs:

docker logs <containerid>

[Smahane]

If looks like there are exceptions due to socket connections. Is it due to python version?
Please see the log attached.
docker.log

[vsoch]

Are you finding the slowness universal, or just for container pulling? The version of python inside the container should be consistent with the one I used (and built into the container) and it looks like there was a timeout on the pull operation. This might either be because:

• the container doesn't exist (so the pull times out)
• something about the pull prevents it from updating the interface.

First let's test the pull - here is how to shell interactively into a container instance. I created a new container, but you could also use your container's id with an exec.

$ docker run -it --entrypoint bash vanessa/tunel
(base) root@5f0703495feb:/code# which python 
/opt/conda/bin/python
(base) root@5f0703495feb:/code# python --version
Python 3.6.4 :: Anaconda, Inc.
(base) root@5f0703495feb:/code# 

And notice above the python version is consistent with yours. Could you share the container you are trying to pull, and where it is from, to get us started?

[Smahane]

Here is the container i was trying to pull:
https://www.singularity-hub.org/collections/783
The slowness is universal. Everything takes too long to load including loading the webpage itself from the start.
I will try the rest of your suggestions very soon. Thank you for being responsive

[Smahane]

So my system's python version is pretty much similar to the container version. I also tried both firefox and chrome. 5 min and the main page is not even loading. Any other ideas please?

bash-4.4$ python --version
Python 3.6.5 :: Anaconda, Inc.
bash-4.4$ docker run -it --entrypoint bash vanessa/tunel
(base) root@2f06e77b7af6:/code# python --version
Python 3.6.4 :: Anaconda, Inc.

Thank you,

[vsoch]

I would try doing a pull from a command line to compare the time. It could also be that you pulled huge images into the Docker container and it's straining system resources, or something like that. Sorry, I can't guess much more than that because I can't reproduce your error - it doesn't work slowly for me.

[Smahane]

The slowness is before getting to the UI or doing anything (loading the home page, clicking a button...) It's not due to the pull. Pulling this image takes 2 seconds via command line.

[vsoch]

I mean the pull within the interface using the tool. When you pull a container in the interface (when it's running) the image goes into the running container...

[vsoch]

You never looked at logs for the container? That would be a logical first step.

[vsoch]

This is an issue with Singularity, or the pull itself. Look at the log:

Traceback (most recent call last):
  File "/usr/local/libexec/singularity/python/import.py", line 109, in <module>
    main()
  File "/usr/local/libexec/singularity/python/import.py", line 86, in main
    layerfile=LAYERFILE)
  File "/usr/local/libexec/singularity/python/docker/main.py", line 85, in IMPORT
    client = DockerApiConnection(image=image, auth=auth)
  File "/usr/local/libexec/singularity/python/docker/api.py", line 92, in __init__
    self.load_image(kwargs['image'])
  File "/usr/local/libexec/singularity/python/docker/api.py", line 146, in load_image
    self.update_token()
  File "/usr/local/libexec/singularity/python/docker/api.py", line 191, in update_token
    response = self.get_tags(return_response=True)
  File "/usr/local/libexec/singularity/python/docker/api.py", line 312, in get_tags
    return_response=return_response)
  File "/usr/local/libexec/singularity/python/base.py", line 288, in get
    return_response=return_response)
  File "/usr/local/libexec/singularity/python/base.py", line 304, in submit_request
    response = safe_urlopen(request)
  File "/usr/local/libexec/singularity/python/base.py", line 152, in safe_urlopen
    return opener.open(url, data=data)
  File "/opt/conda/lib/python3.6/urllib/request.py", line 526, in open
    response = self._open(req, data)
  File "/opt/conda/lib/python3.6/urllib/request.py", line 544, in _open
    '_open', req)
  File "/opt/conda/lib/python3.6/urllib/request.py", line 504, in _call_chain
    result = func(*args)
  File "/opt/conda/lib/python3.6/urllib/request.py", line 1361, in https_open
    context=self._context, check_hostname=self._check_hostname)
  File "/opt/conda/lib/python3.6/urllib/request.py", line 1320, in do_open
    raise URLError(err)
urllib.error.URLError: <urlopen error [Errno 110] Connection timed out>
ERROR: pulling container failed!

Try shelling into the container and pulling the container from the command line.

[vsoch]

What we can try (and it would be helpful if you might) is to install singularity 5.1 (or the latest) in the container (and remove the older version first). If that works for you, I will rebuild tunel and provide it on Docker Hub!

[Smahane]

I did look at the log and here is the one from today where i only opened the page and clicked a button. It doesn't have any errors that are obvious to the slowness.
docker (1).log

Sure i will try that. I assume installing singularity inside the docker container is similar to working inside singularity container? I'm not familiar with docker. I will give it a try

[vsoch]

It could be the bind (see here moby/moby#21485) could you try stopping, removing, and creating a container without binding any volume? I'm sure that this is a Docker issue, logically there is no reason the web interface would be slow - it's Flask with gunicorn (meaning several workers).