hey all, Really appreciate the blog you guys have on the website, it is by far one of the most useful things I've read to get context about this industry as a whole. I'm coming from more of a web/desktop development side, so the industrial stuff is all new to me. I am confused on one aspect - OPC UA vs MQTT, and how to implement a UNS with OPC UA I am collecting tags from PLCs using Modbus TCP using Ignition. I believe that this gets routed to the OPC UA server on ignition - I'd have to use a protocol converter such as https://learn.umh.app/course/our-open-source-docker-container-to-connect-opc-ua-with-the-unified-namespace/ to get MQTT and then I can implement UNS from there? Or am I completely lost? I don't have the Ignition MQTT modules, and those aren't clear about how it works with OPC UA either

See if this article helps at all https://www.linkedin.com/m/pulse/uns-further-explained-matthew-parris-8pjme

With UMH Core you can either directly connect to Modbus devices, or also to the Ignition OPC UA server and then send it out via MQTT. Go to management.umh.app or docs.umh.app and check it out! Has store and forward, data Modeling, etc all in it

The link you posted is to benthos-umh, where we have built now UMH Core around for more easy usage

UMH classic had a very handy redpanda web ui for debugging. What is the recommended way to launch redpanda web ui for UMH Core?

There is no recommended way as for normal users it was more confusing than helpful. If you still want to use it, simply expose the Kafka port in docker run command and then connect to it by hosting redpanda UI on the same host or somewhere else

usability report of Stream Processor editor: topics are not readable, and are truncated even in popups https://cdn.discordapp.com/attachments/984082664678125578/1425491529388265576/image.png?ex=68e7c7f0&is=68e67670&hm=f9705b08115378c4685804d09195551e2ad73b74afcd8c3f0a304ee39cbd3e38& https://cdn.discordapp.com/attachments/984082664678125578/1425491529752903841/image.png?ex=68e7c7f1&is=68e67671&hm=cfc24a05ad7275137cdfce359a6cb0278d7c0e0113cc101bb347286f5177cede&

Another question. I have set up a flow to store separate tags into umh core (uns). Now I have to implement the following logic: when the tag in uns changes -- post a message to mqtt. But the message should be an aggregation of values from multiple uns tags. I thought that Stream Processor will help and created a processor flow with a dedicated data type and contract. Now I see data contract in Tag Browser. I tried using data model topic as input.But it does not return all the contained values -- only one underlying tag that changed. Please give some tips how to create a message containing values of multiple tags

For this use case you can use the standalone flows. Activate the advanced mode in the settings and go to data flows. There you can start with the Kafka to MQTT Bridge template. You can subscribe to multiple topics and do the processing based on any new data point coming in.

thanks, i tried this approach but i am failing at the point to get current values of other tags i need to subscribe to topics A, B, C and when any changes i want to publish json

{
"A": value,
"B": value,
"C": value
}

but the flow triggers when any of A B or C changes and has only one value. How to get other two?

Should the aggregation be over multiple values in the past or just some aggregation of = last_valueA + last_value_B .... ? For the first case, i recommend using a standalone togehter with a memory cache. For a more robust setup, i would suggest using a database e.g Timescale and then in the processing part execute an SQL to the wanted tags you want to aggregate. And for the second option you could use a stream, but be awear that if any value changes in A, B or C the aggregation will get triggerd, so maybe also have to use a standalone for this (memory cache)

i see, thanks for the tip! memory cache looks like an option in my case

processing:
  - branch:
        processors:
          - cache:
              key: ${! meta("umh_topic") }
              operator: get
              resource: last_state
        result_map: root.last_state = this

    - cache:
        key: ${! meta("umh_topic") }
        operator: set
        resource: last_state
        value: ${! json("state") }

With this you could cache the entries, and based on the umh_topic return either null or the aggregation 🙂 https://cdn.discordapp.com/attachments/984082664678125578/1425754541130776648/image.png?ex=68e8bce3&is=68e76b63&hm=14be8e2f56c891cae18a7d0d6873e9d70a48163069d486f568a21997a8637e36&

thanks a lot @krgrsebastian it will take some time for me to set everything up but i will write an update when i have success with that)

If you got any questions on the processing part, just ask the redpanda connect ai. Its based on the redpanda connect docs (e.g benthos :D) and gives you good small answers. It just does no recognise the benthos_umh stuff https://docs.redpanda.com/redpanda-connect/home/ https://cdn.discordapp.com/attachments/984082664678125578/1425757911748972615/image.png?ex=68e8c007&is=68e76e87&hm=d3ca23ede70d677b17e3045a3968469d69f657d77e58de91e4e8612287337ecd&

Woah that's nice, I did not now they had (an actually useful) AI.

Hi All! We’re hiring at UMH to help our customers win with UMH 🙂 If this exites you, openings here: https://umh-systems-gmbh.jobs.personio.de/job/2380318?language=en&display=de - Alex

today I heard a very good trick on this. A IIoT SaaS company puts an LLM in their documentation; As a customer, you can chat with the LLM and ask question about the documentation getting your answer easier. Then using n8n the SaaS company read all the questions asked to the LLM into a rapport and they use that rapport to discuss the roadmap/bugs/difficulties customers have with the product. SaaS said they never got such valuable feedback about their product.

Ohhh, that is a good trick! I love that idea.

We use gitbook, so we can easily enable that LLM as well

This is very much how i use llm. Mainly to read manuals and forums to get answers to bugs and strange implementations. I remember i spent quite some time looking through several manuals for a hardware implementation and could not find a good answer. Then with a correct targeted question to chatgpt it read through several manuals and searched through forums and gave me a detailed instruction on how to solve it, what the problems where with different versions and gave me the sources so i could check for my self. It would have taken me e couple of days to find the info my self.

Also, This is what beckhoff is doing in their new addon which hopefully should soon be released. When i talked to them at the Hannover messe they explained that they have used several different LLMs that had their specific pros and cons and trained them one by one and giving them full access to thier documentation. This is a really smart approach. So if i understood them correct they have like one LLM that helps with hardware(plcs, io, motors, etc) and how to implement. one LLM for help with programming and so on. All connected through a gateway to simplify for the user. I am really eager to test this out when they release it. It it structures like this: https://cdn.discordapp.com/attachments/984082664678125578/1426470885673664563/image.png?ex=68eb5809&is=68ea0689&hm=5f864b5e812431fd4fd96e137cf0903000cb9b80566719ecde174b5b4efa6db5&

FYI, the copy feature is so cool! https://cdn.discordapp.com/attachments/984082664678125578/1429910948591964301/image.png?ex=68f7dbd8&is=68f68a58&hm=1c1151a6badc6680c524cb101069f0f57e3176917632b24386ab26db42cbf9b5&

glad you like it! we are currently evaluating also the @Diederik suggestion on AI for the docs part

Hello all and especially official UNH people here.

I committed to deliver a presentation on the UNS architecture to a data science meetup next week https://www.meetup.com/pydata-cluj-napoca/events/311532017/

I am very tempted to use images from the UMH website because they convey the concepts very well. How can I use the images without getting into any copyright issues?

Hi! Alex here , CEO of UMH. You can use those images for your presentation - no worries. always appreciate if people spread the word. All the best for the Presentation

Thank you very much.

out of personal interest; will the presentation be available online? 🙂