Community Discussions

Hello Cognite experts, I trying to find our some reference material to explore regarding integrating data from RTSP in Cognite. Are there any workflows/POC/documentation/examples available for this data source? 

Hello,Is it possible to host a pickle file (which contains a ML model) in CDF and use it to make predictions for the available data in the platform?If yes, what are the steps ? I browsed the documentation but I haven’t found anything relevant

In some AI demos built on top of Cognite, our approach to real-time inference involves retrieving the latest time series values for the equipment. Since our machine learning models are deployed and running on a cluster outside Cognite, we leverage the SDK to handle real-time inference by retrieving the most recent data as follows:real_time_data_bomb_hfx = cdf_client.time_series.retrieve_latest(id="bomt_hfx_time_series_id")Is anyone working with a different approach to send new data for inference to their machine learning models? I'd be interested in hearing how others are managing this process.We are also exploring more resilient approaches, such as: Streaming Data for Real-Time Inference (Event-Driven Approach): We plan to test Cognite's Kafka extractor as soon as possible to enable more seamless streaming. On-Demand Inference via API: While this approach is synchronous, which has caused challenges for us in the past, we prefer to avoid this method and lean toward more asynchronous

I am making comparisons between time series data in CDF and PI. The reason is that in our tenants the CDF data is not 100% accurate compared to PI.However, from my testing I think that PI performs its aggregations with the timestamps “centered” at the aggregated time periods, while CDF puts the timestamps at the start of each aggregated period. Is it possible to specify how this is done with the Python API? From my study of the docs it appears not to be the case. The same applies to the PI Web API as well: I cannot specify how the timestamps are placed. The agreement with PI becomes significantly better if I place the CDF timestamps at the center of the aggregated time periods.My current workaround is the following:Fetch RAW data from CDF Shift the timestamps by 0.5x of the granularity Resample to the desired granularity Compute mean Interplate any missing valuesThe issue is that fetching raw data is a lot more time consuming than fetching aggregates. I have been playing with fetching

Hi,I am running parallel tasks(Cognite functions) in Cognite workflows. One of the task is creating table in Raw db, where as other one is deleting the table. Whatever table is create, its name is passed as input to other function deleting it.First attempt I ran 150 parallel tasks in a 10 workflows(each 15 tasks). One table was not deleted from raw db  In second attempt I ran 200 parallel task in 10 workflows(each 20 tasks). In this case 8 tables were not deleted from rawI was not able to debug why those table not get deleted. As the delete function status was showing ‘Completed’ on Cognite UI.Can you please help why this tables are not getting deleted? If there is a limit of parallel task execution then how to exceed this limit? And how to debug this scenario where there is no function execution failure?Attaching the test workflow for reference. Run workflow-deploy-job.py file to deploy workflow on cognite project. Edit functions.json and workflow-deploy-job.py files for replacing cog

Hey Guys, I would like to know when will be possible download the Canvas Page.

Ingesting data through OPCUA where the OPCUA Server timezone is set as UTC+8.After ingestion, the timeseries somehow normalize the data to UTC+0.E.g. Ingesting data at 2024-09-11 11:00+08:00But checked the timeseries data as 2024-09-11 11:00+00:00 which mean 2024-09-11 19:00+08:00.How do we ensure that the ingested data uses the correct timezone when storing it?

Hi all.We’re working on defining KPIs to track our digitalization journey, and think that a KPI on the use and adoption of CDF as opposed to “the old way of doing things” would be useful. I’m posting here in an attempt to hear if the CDF customer community can provide inspiration for how to ensure success and good adoption of CDF among users. Does anyone have thoughts to share on their own metrics for CDF usage?My thoughts so far are tracking the number of unique users logging in to Fusion/Charts, number of API calls against CDF, etc. Although I have not found any direct way as a customer to obtain this information. The Sessions API endpoint could possibly be a place to start?And as a follow-up question; CDF usage in itself is a quite indirect metric on the digital maturity of the business, and on acheived value. My ideal goal would be to find a more direct higher level metric for the entire company, which is not as indirect as user log-ins, and also not as narrow as value obtained for

Hey.I am trying to set up a monitoring job in Cognite Charts, with notifications sent to an Microsoft Teams Channel Email. It looks like the email would not be recognized as an valid email.I got this error: “You should select a user or write a valid email address”The email format is xxxx..onmicrosoft.com@emea.teams.ms Any suggestions on how to resolve this?

Can someone tell me about this. if i Want to keep OPC UA realtime data into cdf RAW using opc ua extractor?

Hello I was having the Authenticator setup for logging to zendesk and other cognite sites. But somehow its not there on my mobile now. can you help with re-registering ?my id - nbhatewara@slb.com Neeraj B

2 Questions:What is the user limit for the number of Industrial Canvas pages they can create?Can we auto-delete Canvas pages after some time [Ex: 3 days] to avoid overpopulating the Canvas page/data?

We’re excited to announce that Cognite is now officially listed on the CSA STAR Registry at Level 1! 🌟Cloud Security Alliance (CSA) STAR Registry is a trusted benchmark in cloud security, providing a comprehensive view of the security practices of cloud service providers. Being included in this registry means that Cognite meets the highest standards of security, transparency, and trustworthiness in the cloud industry and joins a competitive list of other recognized leaders. With over 2,000 cloud service providers listed, we’re proud to have our name among the best.So, what does this mean for you? As a Data and AI company, Cognite is dedicated to empowering industries worldwide - from Energy and Process Manufacturing to many other industrial sectors. Our solutions help improve production uptime, optimize operations, and drive innovation. We’re committed to not only delivering cutting-edge technology but also ensuring that your data is handled with the utmost security and care.This reco

Hello everyone.I was testing the offline mode on Infield and got an odd behavior using an iPad, with Safari: when I get offline, I am not able to select any field to insert values nor click any buttons. It is only possible when I am back online.  Is this how it is supposed to work?I have a video but could not attach here,Thank you in advance.

Wondering how to create a faster and more reliable root cause analysis? Check our How-to post:  

We're excited to announce the first bootcamp following a major update! This version now includes the powerful CDF Toolkit. About the CDF BootcampThe Cognite Data Fusion Bootcamp is an instructor-led, 4-5 days course that guides participants through a hands-on example to develop production-ready solutions with Cognite Data Fusion (CDF). While it is possible to grant access and ingest, transform, and contextualize data manually in CDF through the User Interface, creating a proper CI/CD pipeline that allows the development team to promote the project through different environments (e.g. test & prod) is essential in creating a production-ready solution. With the introduction of the CDF Toolkit, it is now possible to replace different types of CLIs with one command-line tool to configure and administer Cognite Data Fusion (CDF) projects. The Toolkit also includes pre-built templates to help you quickly start your CDF project.The Bootcamp has been developed by CDF and DataOps experts and

Hi,I have defined a Data model with two Views, say Employee and Address, both of this do not have any relationship.For instances in Employee and Address view I have kept externalIds same, say emp101, emp102,... for both view instances.EmployeeextrenalId Name Age emp101 Tom 30 emp102 Harry 28  AddressexternalId City Pincode emp101 Pune 123 emp102 Mumbai 456  I am using cognite sdk sync api call to synchronize instances of Employee view.There is one Transformation on Address to update all the pincode values.After running transformation, when sync api is called it always return all the values of Employee instances even if they are not updated. Is there any resolution to this unexpected behavior?  

Is there currently a best practice for adding a geographical location to an object in a data model?I’ve considered simply using a string property to my model containing wkt-formatted strings, or a GeoSpatial feature external id, but none of them seem ideal. I assume there is a size limit on strings - so that wkt might be a bad choice? We’ve previously discussed this for timeseries, where you mentioned that data model geolocation was on the roadmap. Any update on this? 😊 

Hi,I am using Time zone and Calander granularity function (Release timezone and calendar features in DatapointsAPI (beta) by haakonvt · Pull Request #1779 · cognitedata/cognite-sdk-python · GitHub)I am facing an issue wherein I am passing following datadps_lst = client.time_series.data.retrieve_dataframe_in_tz(   external_id=list,   start=datetime(2023, 6, 24, tzinfo=ZoneInfo("America/New_York")),   end=datetime(2024, 6, 27, tzinfo=ZoneInfo("America/New_York")),   aggregates="average",   granularity="1month") For the End date time is given as 27th June 2024; while I am expecting to get a calculation from 1st June- 27th June; I am receiving an output from 1st June- 30th June. the functionality is not considering “end” date while calculating.  

Hi, we are using Cognite PI Extractor and in PI we have time series where there in some cases are only nano seconds between each datapoint. Since CDF use epoch milliseonds how will the extreactor handle such cases? 

Hello,Is there a way to round timestamps of the incoming OPC UA tags to second within the configuration file? I have gone through the documentation but haven't found anything like that. The problem is that for some tags coming from an OPC UA server I make some simple transformations and these tags are coming to CDF with timestamps at ms, whereas the other tags for which I do not apply any transformation on server side, are coming with timestamp at second. The issue I encounter is when I push the data from CDF to PowerBI and pivot the data and I get 2 rows for the same timestamp which PowerBI is displaying at second, 1 row with values for the tags without transformations and the other with the tags transformed which have timestamp in ms on CDF side. I just don't want to apply another query/transformation step to round the timestamp at second on PowerBI side and on CDF this timestamp rounding should be a scheduled transformation running every minute if not every second. Regards,Raluca

Hi Team,I was looking for any limitations to length or any other limitations for Json data type for properties in the data model but was unable to find it in the documentation.Do we have any limitations? If yes, we would want to know. We are planning to use the json data type in the model to store values for an ongoing project as part of our solution.Thanks,Akash@Aditya Kotiyal 

Hi there,Could you confirm if there is a limit on number of tables allowed within single databaseor is there any limit of tables or database per project ?

Hi,Our project uses Cognite Function scheduling to deploy data to CDF, and we struggle to understand what the “Duration” of a Cognite Function actually represents. I have attached a screenshot showing one of our function calls, printing the duration of various function from our source code. As you see the duration of our source code takes 2+ minutes, but the call of the function itself takes less than 2 minutes (start at 7:40 and end at 7:41). Is there a distinction here between what is actually run in the backend, and overhead in terms of instantiating a call to the Cognite Function? Can anyone explain the apparent deviations? Thanks.