BoKKeR

I quickly added the home assistant repo functionality. So far the build is only for x86-64. It should work if you add the repository: https://github.com/BoKKeR/unraidapi-re More info in the readme. I have also enabled issues on the repo. Please open issues there.

Its in the first post in this thread, and in the Github readme. I will add it to the template and the webpage

@emilgil env variable in the template. 20 000 should be 20 seconds Also you are running the tag 6.11 with unraid 6.12. They have to match

No granular way to set polling on each feature, you can slow down polling as a whole by changing MQTTRefreshRate=20000

@Puntoboy what version are you running of unraid? The config folder is the one thats set in the template, probably /mnt/user/docker/Unraid-API

@Puntoboy can you check the config folder if there are any files? You could try wiping them also. You could enable writing the HTML files with the WRITE_HTML_OUTPUT=true environmental variable. That will generate more logs about the scraping. You can share these logs privately with me for more debugging.

@corgan I am not sure how mqttDisabledDevices.json works, It looks like you can disable docker for a full host, but not really specific containers? I could look into disabling auto discovery since its quite messy with the docker containers... I am also not sure if the discovery feature is broken. I have not touched it from the old project and frankly I am not sure how that feature should behave. I will have to do some research or you can fill me in on the details if you have experience. The update intervall can be possibly adjusted with an environmental variable `MQTTRefreshRate` also seen on the screenshot. @viktortras I have not tested with a public myunraid address. Could you try locally on the same network? @Orishas Noted, will have to revisit this issue @vicious-transgression6851 You seem to have all the data so it might be unrelated to unraid-api? I mostly use this with home assistant. @Puntoboy There is no home assistant version. I am using it as a container deployed on UNRAID connecting it to HA with MQTT. The home assistant version would just run the unraid api container within HA OS.

Yes, when I find time i will the entities that are broken. Hopefully this weekend.

I am afraid that the current plugins and docker container community supporting unraid are not versioned properly to support a big influx of users on old unraid versions. There is no easy way of installing a plug-in which matches your old unraid version as far I know. The way of making these components match is just updating to latest bleeding edge of unraid OS and updating all supporting plugins/containers. With this change there will be many users stuck on older versions. They would have to manually match plugins to unraid versions. I maintain the unraid-api and there is not enough bandwidth to backport features. I can only implement those for the newest unraid versions. I keep tags for older releases. I am unsure if plugins have a similar mechanism. I welcome the change but I am afraid it will create a lot of noise for individual contributors who maintain smaller projects related to unraid. Also is there an official release date ?

After the files from @light2089 I have fixed so the version returns an empty string in the meanwhile if the version is not found. I will investigate further since it works on my test machines and I suspect its some kind of a custom setting which hides the version on the main page

@BlaM I would like to add this also, but unfortunately I dont own a UPS, is there a way I can emulate a UPS? Since once a feature is implemented it has to be maintained going forward. @light2089 Thanks for finding this bug and opening a PR (I somehow didnt get a github notification). I cant reproduce it on my 2 systems running 6.12.6, Are you using any custom theme? Could you run the container with WRITE_HTML_OUTPUT=true and send me over the dashboard.html file from your /config folder in private? So I can find a version a code that works for all of us. @ABEIDO The logs are coming from unraid itself. Since the container needs to ping the dashboard there is no way this container can fix it. The unraid team might be able to help in the future how to surpress them. I will look into the HA problem

I have ended up using the terminal/ssh. example command: docker logs --follow `docker ps -aqf "name=Unraid-API-RE"`

I have seen some reports of it working and not working on 6.12.6, I am currently away from my computer so I can’t test it but please submit your dashboard html files after running the container with WRITE_HTML_OUTPUT=true and send them over to me in a PM since they are not anonymised. @k11su @mkono87 @chompy18 try running npm ci instead, alternatively check how the docker file is setup. The packages are broken since before I took over the fork I will have to revisit the Unraid dashboard getting slow once I find more time to write a backing off strategy

@syper @ByteHeaven could you try using the MQTTRefreshRate=60 env variable. Setting the refresh to every 60 seconds or even 120. Instead of the default 20 seconds. Also close the website since I noticed that the web browser has its own 3 second times, each tab would start a new 3s process. I will have to rewrite that... I suspect that when the server/ui becomes slow the unraid-api does not back off and just queues up more tasks ending up slowing down the UI indefinitely. I would like to rewrite that whole logic but my current fix would require an external redis service. Or implement some sort of logic for exponential backoff https://stackoverflow.com/a/52772266/2515404 @skank Does the entity show up as following?

Are you able to wipe the /config folder and start again?

@apedance 6.9-2 is not supported, update to 6.12 for maximum support

Keep the variable false if it works

Did you update to the naming fork? @Blade007 Are you able to add WRITE_HTML_OUTPUT=true to your env variables, it should log every time the site gets scraped. Additionally you could try to remove the config and start fresh. About the memory leaking @Foehammer1982. I am not saying that its not happening but I cant seem to reproduce it on my server (last 7 days view)

I have potentially fixed the naming problem. Do try out this branch bokker/unraidapi-re:6.12-naming I have also grouped the VM usb devices under the VM There is also an option to set icons in the MQTT messages for entities. I have not had time to choose any. https://pictogrammers.com/library/mdi/

Interesting idea but I dont think its as clear cut as you might think. You will have to come up with a common structure that the url would have to supply. Or alternatively add a little script inside of unraid-api where you could transform the data to the desired output. Its quite a lot of work currently for me. As always, contributions are welcome. You can probably find me on discord for discussions

@Foehammer1982 I ran some script to detect potential node leaks and it triggered right away. I am not sure if its false positives, will have to investigate more. The whole polling is done with multiple setTimeout calls so if anything thats my suspect. Also I am questioning the need for multiple server support, now with the MQTT change. Considering that users could just run multiple docker instances for multiple servers and it would possibly solve some of the problems around MQTT naming. The list of issues currently are looking into are: - new home assistant MQTT entity names - server shut down ends with ON state - HA entities go unavailable on HA restart (can be fixed with retain:on but those messages have to be cleaned up) - memory leak - make the image support multiple unraid releases with single docker image - deleted dockers end up on HA (unconfirmed)

sorry, I have been mostly focused on 6.12, please update if you are able to. Going forward I will try to support older OS. The problem is that the project needed so much rewrite to get to a good shape. And the work is still not done. To be able to do that in two branches turned out to be much harder than expected. Currently I am focusing on fixing up the codebase in 6.12

Could you tell me more how you detected the memory leak? What was your setup? @lanky8804 did you manage to figure it out? Could you try removing the info config file and starting fresh? there is also an env variable that might show if the scraping is working: WRITE_HTML_OUTPUT=true

Thanks I fixed that typo!

Good news everyone! I located the naming issues cause https://community.home-assistant.io/t/psa-mqtt-name-changes-in-2023-8/598099 I couldnt reproduce it since I was running the older version of HA. Here is an example error you might have noticed in HA: 2023-10-03 10:48:52.147 WARNING (MainThread) [homeassistant.components.mqtt.mixins] MQTT entity name starts with the device name in your config {'payload_available': 'True', 'payload_not_available': 'False', 'json_attributes_topic': 'homeassistant/unraidvm/home-assistant-container', 'name': 'unraidvm_docker_home-assistant-container_restart', 'unique_id': 'unraidvm_home-assistant-container_restart', 'payload_press': 'restart', 'device': {'identifiers': ['unraidvm'], 'name': 'unraidvm', 'manufacturer': 'ASUS ASUSTeK COMPUTER INC. , Version Rev 2802', 'model': 'Docker', 'connections': []}, 'command_topic': 'homeassistant/unraidvm/home-assistant-container/dockerState', 'encoding': 'utf-8', 'availability_mode': 'latest', 'qos': 0, 'enabled_by_default': True, 'retain': False}, this is not expected. Please correct your configuration. The device name prefix will be stripped off the entity name and becomes '_docker_home-assistant-container_restart' Meaning the friendly name cant start with the same string as the device name! I will have to look into proper naming with the new HA release. I am open to suggestions!