bpwwer

Known issue with IoX 5.5.5 and PG3x, it's being worked on, but no work-around yet.

3 and 4 happen are what PG3x does to migrate the node servers over. The PG3x backup/restore screen has two options for restore: 1) Restore Backup 2) Migrate from PG3 Backup Using #1 won't do anything with a PG3 backup since the backup formats are different between PG3 and PG3x. Using #2 looks at each node server in the PG3 backup and one-by-one migrates to the PG3x, you should see status updates on the PG3x screen as this happens. It's not quick, as I was testing with 30 node servers or so, I would walk way and come back later since it can take a minute for each node server. When the migration is complete, the node servers should not be "unmanaged" state. It won't hurt anything to the do the migration again. It doesn't really care what the current state on PG3x is, it will start clean and migrate from the PG3 backup.

Those are instructions for migrating from a i994 to either Polisy or eisy. Not instructions for migrating from Polisy to eisy. while some the steps may be the same, it doesn't really cover the PG3 to PG3x migration steps I specified above.

What instructions did you follow? "Unmanaged" means that the node server config exists on the IoX but the config points to a different instance of PG3(x). If you simply restored a Polisy backup on the eisy, this would be the result. It would restore the Polisy's node server config so all the node server configs would be pointing back to the Polisy's PG3 and would still be considered managed by the Polisy's PG3 instance. To migrate the node servers, you need to backup PG3 on the Polisy and then use PG3x's migrate from PG3 backup button to migrate the PG3 backup and convert it to PG3x format and change the node server configs to point to PG3x.

While trying to track it down I was restarting my PG3x very frequently and it would tend to be the same node server that it would get stuck on for a while. Initially it was YoLink, then it switched to WeatherFlow, then it switched back to YoLink. So I'd say there's a 50/50 chance that restarting PG3x will result in the Elk node server not starting.

@DennisCThis looks like the same issue we've been trying to debug for the past couple of days. One (or two) node servers don't start and won't start manually with a "already starting" response. Last night I managed to track the issue to the service that PG3x uses to start node servers. PG3x calls the service to start the node server and .... nothing, crickets. the service never starts the node server and never responds back to PG3x so PG3x is sitting there waiting for it. It's seems to be somewhat random as which node server(s) are effected and it can change when PG3x is restarted. For some folks it's been the YoLink node server, for others the Elk, For me, it switches between YoLink and WeatherFlow. I don't have access to the source for that service so right now I'm waiting to here back from the person that does.

Very likely the upgrade wasn't finished.' With the eisy, there's no good way to know when it has finished and PG3x may start before some of the other services that it requires. In this case, it looks like the MQTT service wasn't initialized when ELK first tried to start. The latest update seems to be causing quite a bit of trouble.

There's another thread on this but this might be a better location for it. I just did an upgrade to upgrade my IoX from 5.5.4 to 5.5.5. My PG3x was already running the latest version so it didn't get upgraded. I have 7 node servers installed but not all are enabled. I hadn't really paid much attention to which were running and which weren't but now that I am, I am seeing this issue: After the upgrade/restart YoLink was not connected and I was able to reproduce the problem. Nothing looked wrong in the log. After restarting again, YoLink started fine but my WeatherFlow node server was now in this state, disconnected and already starting when trying to start.

Thanks! I'll take a log a the log. No log showing for the node server or no log available? Unfortunately, with PG3x those are no longer the same thing. With PG3x, it doesn't display the existing entries when you select log, it only shows new entries that happen after you select log. So the node server starting log entries may be in the log file, but not displayed, which is why I requested the log package for it.

I installed and did some testing with the Occupancy node server and I'm not seeing any issues. There is at least on support ticket open that sounds like it might be a similar issue. PG3x shouldn't have any interaction with the Occupancy node server other than to show an Unmanaged dashboard entry for it. I've verified that PG3x isn't attempting to delete or modify it in any way. When you say it disappears, where does it disappear from? From the admin consoles node server list? Or just from the PG3x's dashboard? or both? Is there something that triggers this? I.E. when you restart (IoX, PG3x, the eisy) or is it random?

@dwengrovitz @sjenkins can one (or both) of you reproduce the error and then PM me a log package for the YoLink node server? I just installed it and as far as I can tell, it's working fine. But I can't configure since I don't have anything it would work with.

It looks like it's getting stuck trying to start the node server. We were having problems with people pressing the start button multiple times so it call start while it was already trying to start. So now it aborts any attempt to start the node server if the start process is already running. It appears that something is going wrong while trying to start YoLink where it's not starting but also not failing and is stuck in a limbo state. I'll look into it.

Did you check the node server log? It should provide a bit more details on the error. Are you using the eisy's wireless network for your network connection? If so, the discovery process may not work with it. If you can, switch to a wired connection and see if that works.

Most likely the browser has the version number cached and it's the browser page that needs refreshing.

That likely means that the node server never stopped when PG3x told it to. If it's running, you can't start a duplicate. It also could be confused so the first step would be to do a stop, wait 5-10 seconds and then try starting it. If that doesn't work, you may have to reboot the eisy.

Yeah, I agree that it is strange. They should match. Possibly it's a bug in the library that's being used to do the actual queries. It is a reverse engineered library as the Emporia cloud API is not publicly available.

I fully agree with this. At this point, I don't really feel that's is my place to announce a change of ownership and that it's something the original author should do. Unfortunately, UDI doesn't have the resources to police the node server store so it's really left to the individual developers to be upfront with the level of support they're capable of providing. It's not hard for a developer to update the description or readme for their node server to say they are no longer going to be providing support. But this is where having ratings/reviews could be really useful. Maybe adding a support commitment level would be useful, but that again would require the author to update it if their situation changes. This discussion has been good in that it's given me a lot to think about for the node server details. Thanks @palaymanfor initiating this.

Why does it matter? Are you just curios? I'm reluctant to name the developer (which would be obvious by listing the node servers). If you feel you have a legitimate need to know, it might be better to move this to a PM. My point was that there are ways to properly transfer support of a node server should the developer feel it's something they can no longer support.

I suspect that most node servers are under the MIT license as that's what the template and examples use. So my guess is that most just copy the same licence file over and don't change. However, I can really only speak for my own node servers. Those that I've written and sell are licensed under a proprietary license that prohibits modifying or re-distributing the code.

The monthly total comes directly from the Emporia cloud service. The node server doesn't do anything to change what the cloud service sends.

Most node servers are written in Python. The tools needed to write them are available on both Polisy/eisy. Which is mainly and editor and the Python interpreter. There are also Python IDE's available like pycharm but those would typically need to be run on something other than the Polisy or eisy. It is also possible to write node servers in Node.js and again this can be done directly on the Polisy or eisy using an editor and the node interpreter. The environment can be as simple as ssh'ing to your Polisy/eisy as admin, create a directory and add/write the files needed to for the node server. For Developers, PG3 can then install your node server from that directory for testing and continued development. There are many more advanced ways to set up a development environment. So folks use a Windows environment and mount the Polsy/eisy file system to the Windows box using NFS or Samba and can then run an IDE like pycharm on the Windows box. There are example node servers in the non-production store that can be used as reference and there's also a template node server that can be built and run. PG3 includes some help files on how to build and package up node servers to submit them to the node server store, those help files show up to registered developers. registered developers also get access to a Slack channel for developers. That's where we help each other out.

@bigDvetteI added it to my list of things to do. It sounds like it could be a bug in the node server verification code, but I have't had a chance to look at it yet.

Good point. In most cases, I don't believe it is, but it should be. Another thing to add to me todo list.

If the node server is proprietary and doesn't grant permissions to modify the code, then yes, creating a new one is the obvious answer. But again, this depends on what the node server developer want to do. Recently one of the developers wanted to stop supporting their node servers and passed them on to someone else to continue supporting them. Most of the PG2 node servers that were abandoned, were ported to PG3 by me since the license allowed for that.

No, that would be a bad assumption. Developers can make the source available or not, depending on how they license the node server. Many free node server do have a fairly permissive license and do make the source code available. And no also to the second question. The license the node server is released under is determined by the node server developer. It would be illegal for a third party to release the source if the developer doesn't allow it. The node server ecosystem isn't really large enough at this point to attract corporate developers so today it's people creating and supporting node servers in their spare time. So of course there's a concern that they might stop supporting their node servers.