Launched: Data Science Portal


#125

Any chance you got a !include_dir_named is not supported workaround?


#126

Hi @bachya, thank you very much for your efforts!
Unfortunately I get the following message every time I try to pull your Docker Image:
ERROR: pull access denied for bachya/jupyterlab, repository does not exist or may require 'docker login'
However, I have already entered my correct Docker Hub login data via 'docker login'.


#127

The compose file builds it based on the dockerfile in ./jupyterlab/docker

for a premade one, see this post further up.


#128

rpi of armhf compatible images ?


#129

My 2 cents on the topic.

A lot of people seem to be getting confused between Hassio and HassOS.
HassOS is a premade image of an OS with the bare necessities to run hassio supervisor and HA docker container…

Hassio itself, the homeassistant environment running via docker containers, in my opinion is definitely the way forward for homeassistant, and is fully justifiable as being the recommended installation method.

At work, docker and containerisation in general is something i’ve been working on implementing at various levels.
The simplicity of migrating to better hardware, to replicate instances, or add in load balancing, is phenominal.
In general, that is the way that enterprise IT is going as a whole, and should be what developers try and steer their mindset towards on any projects starting these days.

Yes, if you’re developing new components and such then it can be beneficial to test on bare metal machines. Or, if you’ve a very specific setup that can’t run in containers then fine, do it that way. But to have the gall to comment on an announcement that was intended to make things easier for people who are running the recommended install method, complaining that you haven’t got written instructions for your very specific setup, I frankly find that atrocious behaviour.
If you’re competent enough to go vanilla, google and set it up yourself. It’s less an integration, so shouldn’t be something HA developers should be holding your hand through.
Better yet, write a script to install and set it up, and submit a PR!

A lot of people just need to spend some time learning what docker is, rethink why you have your setup the way it is, and check out the pro’s and cons of staying bare metal.

It’s about to turn 2019. Get with the times and move forward.


#130

At the beginning of this thread, when @ReneTode said:

i think its a very bad sign if there are things created for home assistant that are not usable for not hassio users

people jumped in to say that he was overreacting. In contrast, 90% of your post serves to confirm his concerns.

If Home Assistant’s core developers are on the same page as you, then Paulus ought to issue a statement to that effect, namely Hass.io is the way of the future for Home Assistant. This would help minimize needlessly contentious threads like this one (which borders on a ‘religious war’ between factions).

  • If it’s official policy then non-Hass.io users will understand their choice of installation method is now being relegated to the dustbin of history. We shed a collective tear (/s) and move on to Hass.io.
  • If it’s not official policy, then there’s no reason for non-Hass.io users to be pilloried for voicing their concerns (or displeasure).

… to have the gall to comment on an announcement … complaining that you haven’t got written instructions for your very specific setup, I frankly find that atrocious behaviour.

The so-called ‘very specific setup’, happens to be the way everyone installed Home Assistant for many years. Hass.io is a recent development (and a welcome one) so to refer to the traditional way of installation as being a ‘very specific setup’ is disingenuous.

I recall a podcast, from several versions ago, where Paulus commented on the installation statistics being collected. There was almost an even split between Hass.io and non-Hass.io users, so it’s hardly ‘a very specific setup’.

If you’re competent enough to go vanilla, google and set it up yourself. It’s less an integration, so shouldn’t be something HA developers should be holding your hand through.

The implication here is that by following the well-documented alternative instructions for installing Home Assistant, one is now deemed sufficiently competent to no longer require any further assistance for anything else. That’s unrealistic.

Getting past the low-bar of ‘alternate installation’ is no guarantee one can handle all future situations with merely “google and set it up yourself”. The proof is in the fact this forum is full of posts made by non-Hass.io users asking for help on a variety of subjects.


#131

Not to mention the HUGE number of Hass.io users who come on here asking questions how to do a lot of mundane things because they’ve literally never had to figure ANYTHING out in HA themselves. I’d venture to say that most (if not the large majority) of the questions are answered by the people who either currently don’t use Hass.io or those who used to not use Hass.io but have decided to make the switch.

And, relatedly, to reply to the person above who stated that it’s only .0001% of users complaining so the devs can just ignore them - Have you actually looked at the threads on the forum where that “.0001%” contributes a HUGE amount of knowledge. I’m on this forum a lot and I constantly see the same names here, “complaining” as you say, helping a huge number of people who don’t have a CLUE how to fix something when it doesn’t go EXACTLY as the instructions say or who want to create some special template to do a specific uncommon task. That knowledge comes from having to slog thru the old stuff and figuring it out with a big dose of “prior knowledge” from other people who also had been around even longer. It’s not like, as we all know too well, that the documentation is a vast fountain of information.

I personally believe that it more takes gall to say that than point out a concerning path that seemed to be implied by a poorly thought out announcement.


#132

I should probably apologise for the post before. Ranting without rereading back through, and being a bit too harsh.

But, in hindsight, HA should probably get to a point where a decision is made as to what the “official install method” should be, and announce it as so if it is Hass.io


#133

Hi @kylerw,

thank you very much for your preconfigured container.
But every time I try to create it I get this in the container logs and the container won’t start:
PermissionError: [Errno 13] Permission denied: ‘/home/jovyan/.local’,
Container must be run with group “users” to update files,
I also tried to gave recursive 0777 rights to the /home/jovyan folder, but won’t work.


#134

I have to argue that nobody has my same problem. Sorry for my repeated request, but I’ll really appreciated who has some good ideas to overcome (or better understand( my problem.
Thank you again

Pls, tell me if I’ wrong to ask my question inside this thread.


#135

Hi,

I, as others, am curious about the relation/comparison with influxdb+grafana, specially for long term data support. I am interested in keeping a log and analyse power and heating usage for last years. I can see the potencial of using directly python analysis tools instead of grafana for this. However, I don’t know how to store all the data. As far as I know local .db default is to keep data for 10 days. What would be the recommended course of action to use this over long term periods?

  • Just increase the default 10 days to years (huge db problem?)?
  • Switch from local db to postgresql or mysql?
  • Keep backups of local db and manually merge the needed data later?
  • Keep influxdb and somehow make this data tool talk directly to influxdb (if it makes sense)?

regards,


#136

mapping your docker volume to gets around the permissions to starting the docker instance. However it still has permission issues inside workspace using below:

path-to-dir:/home/jovyan/workspace


#137

Running HomeAssistant in a docker container does not necessarily mean running hass.io. I run HomeAssistant in a docker container. I do not run hass.io.


#138

Thank you so much for your advice.
But I think I will try to realize some of the notebooks from @robmarkcole with Apache Zeppelin and Docker.
I use that myself partly at work, so I know it better and it seems to be a bit more accessible to me from the maintenance and handling side.

Of course it is not as versatile and powerful as jupyter (atm).


#139

After abit more investigating into docker compose users running into permission issues whilst mounting your directory. Below is a working compose with the appropriate environment variables configured for the jovyan user. Adding the following will rectify the issue:

  jupyter:
    container_name: jupyter
    restart: unless-stopped
    user: root
    image: kylerw/hass-data-detective
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - /your-path:/home/jovyan
      - /your-path:/hass-config:ro
    environment:
      JUPYTER_ENABLE_LAB: 'yes'  
      NB_USER: jovyan
      NB_GID: 1000
      NB_UID: 1000
      CHOWN_HOME: 'yes'
      CHOWN_HOME_OPTS: -R
    ports:
      - "8888:8888"

#140

Didn’t try it yet, but that is really a great great idea, because history is just unreadable when you have more than 10 sensor and devices.

i like all your new ideas


#141

This is an awesome project. I’m sure all the data HA store are not fully exploited. It would be nice to have a concrete exemple on how Jupyter can help to create an automation according to our daily routine.


#142

got the following error running this docker on raspberry pi3 - seems the image is not compatible with armhf?
Linux 4.14.79-v7+ #1159 SMP Sun Nov 4 17:50:20 GMT 2018 armv7l GNU/Linux

[email protected]:/opt/docker/jupyterlab# docker-compose up

Creating jupyter ... done
Attaching to jupyter
jupyter    | standard_init_linux.go:190: exec user process caused "exec format error"

#143

You might need to build your own compatible dockerfile to support your platform. Heres a git that hasn’t been updated in awhile but can guide you on what is needed for a rpi version.


#144

I have never commented on a Hass blog post before and I haven’t been logged in to here in ages. But I eagerly await every single release and opportunity to read what is new with this amazing software and community. II like the idea of the data science portal but perhaps making it clear in a post announcing a new feature what is required of hassio and non-hassio users in order to benefit from it. Just simply saying “non-hassio users will have to install and configure additional elements in order to benefit from this new feature” outright in the OP would have killed 90% of the negative feedback I’ve seen in this thread. Instead, devs appeared to dig in behind the “well you should be able to figure it out as a more savy user” defense. This is poor customer relations practice in a community that touts its inclusive, friendly, and supportive atmosphere. Case in point: the first response I read/saw from Balloob included a copypasta of the definition of the word “portal.”

I don’t know what else to call that other than a dick move.

Feels like you just LMGTFY’d the entire non-hassio userbase.

I’m not actually all that upset here. I just see the reasons why others were and wanted to voice my opinion that the blowback from hass devs was less than cordial and as such is a marked difference from dev responses would have been worded in the past. If there is warning to be headed here, it is change in tone from the leaders of this community.

Still love Home Assistant though. #OxfordCommasNeverDie

In case one is curious…