Advice reqeusted to migrate MSR from Bare Metal to Container

[toggledbits]

We're 60 days out from the 1.0 release date I gave earlier in this category. For about the next 30 days, I'll entertain new feature requests for anything you all feel is absolutely needed and not yet present for a 1.0 release, and you're going to have to make a strong case. Short of that, it will have to live on the wish list for a possible future release.

Identify and discuss in this topic, please... go!

[LibraSun]

The only thing that comes to mind is fleshing out the Docs (gonna be a fun chore!) and introducing more little (i) icons throughout the UI so that users can quickly jump to the related section of Docs for that panel/pane/action.

Right now, 21140 is like butter and I have no pressing outstanding Feature Requests to suggest.
Thanks for all the hard work you've put into making MSR as awesome as it is, even in beta!

[Matohl]

Virtual Devices as discussed here:
https://smarthome.community/topic/546/let-s-talk-about-msr-provided-virtual-devices-not

[Elcid]

possibly add future wind readings to weather entity like rain(1h and 24hr)

[MikeReadington]

+1 on the docs. As a non-programmer, I struggled with what could probably best be described as the lack of prerequisite programming knowledge. I certainly would never expect a programming course in the docs, but examples in the "how-to" section will really help point people like me in the right direction. The installation (docker), logic, and concept sections were excellent for me, but I already had the basic gist of things from Vera Reactor.

+1 on the virtual devices.

Dashboard: I don't know if you view this as a separate but related project, but since MSR concentrates entities in one system, it would be a wonderful addition.

The only other thing I had thought of was a global "pause" button. I had to make lots of controller/device changes at one point, and I found myself looking for a way to put everything on hold while I reconfigured.

I just wanted to say thank you again for everything you have created. MSR has brought a level of maturity to my automation that was not previously possible for me to accomplish at my skill level.

[toggledbits]

Potential change: now that constraints apply to groups in a Rule's reactions, I see no value in keeping the Constraints as a separate subsection at the top level of a rule (that is, the section between Triggers and the Set Reaction when you edit a Rule). So, I'm thinking about removing them from here (any rules you have with constraints in this area would just get them restructured into a group constraint).

Thoughts?

[MikeReadington]

On the constraints, once groups were introduced, I stopped using the constraint subsection unless it was a straightforward rule that did not require groups. I think opinions on this would vary based on the way people have settled in on rule construction and layout. Groups allowed me to consolidate my rule sets, so I have less use for them.

I thought of one other thing that would be helpful to me personally...

Would you or anyone else see any value in an "entity in use by" inspector/report generator? As things got more complex, I found myself doing a little digging when I needed to address a logic issue with an entity used in multiple sections of logic.

[LibraSun]

+1 on "Entity in use by" (or "watched by" or "referenced in") and see that as a possible extension of the objects we already see under ENTITIES. If, for instance, "Sofa Lamp" was referenced twice in the "Dim Lights" rule, it could list both occurrences as:
.references: Dim_Lights._Set_Reaction, Dim_Lights.Expr_DimLevel, where both are clickable links directly to those subsections of the referencing rule.

[MikeReadington]

This post is deleted!

[MikeReadington]

+1 Libra. A clickable link to take you to the location of use would be really nice.

[LibraSun]

@mikereadington said in Version 1.0 Pre-release Discussion:

+1 Libra. A clickable link to take you to the location of use would be really nice.

Although we have to acknowledge that expressions are able to reference entities without overtly looking like that's what they do - such as using loops and arrays - so it might be a technical challenge for MSR to pick up every single reference.

[cw-kid]

@toggledbits said in Version 1.0 Pre-release Discussion:

now that constraints apply to groups in a Rule's reactions, I see no value in keeping the Constraints as a separate subsection at the top level of a rule (that is, the section between Triggers and the Set Reaction when you edit a Rule). So, I'm thinking about removing them from here

Really?

Are these not two separate things with different uses.

I use the Contraints section of a rule as the main set of constraints / conditions for that rules Reaction / actions being run or not.

And then in some of my rules in its Set Reaction I also have further additional and different "Group Constraints" that are only evaluated and carried out if the Set Reaction actions were even run in the first place by the main constraints being met.

Personally I do not like at all this idea to remove the rules Contraints.

[cw-kid]

I have to strongly object to removing the rules Contraints section, which would be a major change to the concept of a rule and its work flow.

A logic engine has triggers then conditions then actions.

This is the same concept PLEG had and I used it for a decade.

We then asked you to give us the ability to have further additional conditions within the actions chain and you gave us that with Group Constraints.

The system is perfect as is.

What you are suggesting is to have triggers then actions with conditions mixed in with the actions.

Keep the main conditions in the separate Constraints section as they are now.

It's your project and creation and you can do with it what you like, but I won't be happy with this proposed major change if it happens.

[SweetGenius]

@toggledbits , As you stated in another thread "Capabilities on Hubitat are free-form, and there's no indication which capability might be the most useful or primary". Could you add the ability for a user to manually select the primary capability for each Hubitat device? This would allow the Entities screen to display info in the primary value. If the Dashboard feature is added in the future then the Hubitat device should display correctly there also.( I also would like to request the dashboard)

[Fanan]

This might be off topic and it also reveals my lack of programming skills. But I wounder how is MSR going to be installed, once it leaves beta? I use it on a stationary Windows 10, and the problem is if the power goes out. Then my PC reboots and I have to start the app and change catalog and finally execute "node app.js". I could install it on a RPi, don't know if it can work alongside and on the same RPi as I got Home assistant installed? Then it would start straight away, when the power gets back. To summarize, it's basicly two questions;

1. In which ways are we going to be able to install and start up MSR?
2. Is it possible to install MSR on a RPi that already is running HA? Or on a Home assistant blue (Odroid N2+)?

Another feature that would be nice is to have a optional setting of automatic updates.
Regardsless - MSR is a great (one of the best) contribution to the smart home community - at least for me. Great job!!

[Matteburk]

@fanan The same thing happens with my RPI4 ... I have to start it manually...

[cw-kid]

@matteburk

There was a thread about having MSR auto start on a Pi. I set it up on my Pi and it did work MSR was started automatically after a reboot / power on.

https://smarthome.community/topic/439/how-to-auto-start-on-rpi-reboot?_=1621588014603

[therealdb]

@fanan installing on a poi side by side with anything else will not hurt. On windows, just schedule it to run at startup via task scheduler.

[toggledbits]

@fanan At least initially, like many other projects, I expect MSR to be delivered in the following forms:

1. Primary: Docker container. This facilitates not only a simpler integrated package to install that contains its dependencies, but also easily takes care of starting at boot on most systems. There are currently Docker containers available for Linux amd64 and armv7l (i.e. RPi4) architectures.
2. Generic tarball. This will be a manual install for platform power users (Linux, Windows, and anywhere else nodejs and the companion dependencies can be installed by the user). There are available examples for automatic startup for systemd.

There's no reason you cannot install MSR next to Hass or anything else on a platform that allows you to install such things (clearly you're not going to install it on your Hubitat). I don't know anything about Hass Blue, but if it lets you install docker containers or the generic package, I see no reason you can't run them side-by-side.

@cw-kid said in Version 1.0 Pre-release Discussion:

@matteburk
There was a thread about having MSR auto start on a Pi.

Slightly modernized version of the same here, should work on any platform that supports systemd: https://smarthome.community/post/8011

[toggledbits]

@fanan said in Version 1.0 Pre-release Discussion:

Another feature that would be nice is to have a optional setting of automatic updates.

I'll address this issue separately.

I think there are two interpretations here, so I'll start with the interpretation that I would more specifically call "unattended upgrades". Philosophically I'm opposed to unattended upgrades in all things, particularly on things that become "mission critical" in my environment. There is no "good time" for such updates except when I'm there to initiate them and recover if they come a gutser. History has shown that they otherwise will happen when I am far from home, usually for a sustained period and while Internet access is unavailable or inconvenient. When fully unattended upgrades to MSR happen, that will be your sign that I am no longer associated with this project.

The other interpretation is "push-to-start" upgrades from the UI. This is, frankly, a very low priority for me, and will not be something I entertain for 1.0, or likely 1.x. Docker containers are easily upgraded with a four-line script on command line-driven platforms, and still pretty painless in, for example, the Synology GUI (and still doable on the command line there as well). For users of the generic tarball, the detar and restart is even simpler. I understand that one click from the GUI would be easier, I don't disagree there, but keep in mind you've all been working in a mode where, for much of the history of its availability, MSR updates have come almost daily. That is not the norm, and I would not want it to be in production. In fact, production will go more to the model I use for R4V, where there is basically a very slow-moving release channel, a faster "stable" channel, and a bleeding-edge "latest" (what you have today).

The feature also introduces something that Reactor currently doesn't (on its own) require: access to a cloud infrastructure. Upgrades from within MSR require it to access cloud services, some existing, some not. This implies a lot of things, not the least of which include its security, its construction, and its cost to maintain and serve. If it gets done at some point, it will not come lightly, or easily, and it will likely be accompanied by license fees.