Model deployment failed on connector TW.AnalysisServices.ThingPredictorConnector

[darcudarcu]

Hi,

I have the ThingWorx 8.1.2-b51, it is not a trial or student license.

I can upload datasets and create models but at the moment I cannot publish them. I get the error: Model deployment failed on connector TW.AnalysisServices.ThingPredictorConnector.

I have tried to change the predictive-standalone.yml and redeploy the ThingPredictor JAR file from the command line.

In command line (running the JAR) I get warnings like: 

log4j: WARN No appenders could be found for logger.

log4j:WARN Please initialize the log4j system properly.

Cannot resolve place holder for ${microservice.server}

And different warnings about varaibles (ex: data.service.url.prefix) being set twice.

I believe there was a previous installation of the PredictorConnector, maybe uninstall the previous one might solve the problem? How can I do that?

Kind regards.

[cmorfin]

Hi @darcudarcu 

Is there any chance you could use a newer version ?

Release 8.1 has reached its end of life last year (see https://support.ptc.com/cs/product_calendar/PTC_Product_Calendar.htm#RANGE!A529)

Thanks

Christophe

[darcudarcu]

Hi @cmorfin , 

if that is the only way to solve the issue I will take that possibility into account.

Are there some work-arounds to avoid (at least temporarily) the upgrade?

Regards

[cmorfin]

Hi @darcudarcu 

Usually those error were due to config issues. So likely problem in properties or yml file.

You can upload the following:

- system-environment.properties file

- predictive-standalone.yml

- the exact command you use to start ThingPredictor

we'll see if we see something obvious there, otherwise the upgrade will probably be needed.

Kind regards

Christophe

[darcudarcu]

Hi @cmorfin ,

the command string is: 

java "-Dvariables.file=file:///C:/xxx/WORKFILES/WVER/ThingWorx-Analytics-ThingPredictor/system-environment-variable.properties" -jar thingpredictor.jar server predictive-standalone.yml

The files are converted to .txt because of uploads issues, normally they have the standard extension.

[cmorfin]

Hi @darcudarcu 

Can you add the following line in your system-environment-variable.properties file:

microservices.server=localhost

Then start again the ThingPredictor agent.

Kind regards

Christophe

[darcudarcu]

Hi @cmorfin ,

it fails to connect to the EndPoint (Error from: Command Prompt). Then I guess I have to try to Upgrade...

Are my things and Mashups at risk if I upgrade?

[cmorfin]

Hi @darcudarcu 

Could you copy the content of the ThingPredictor log into a file and upload it here ?

Thanks

Christophe

[darcudarcu]

Hi @cmorfin ,

hereby there is the log.  At the beginning there are different issues than in the end, since afterwards I modified the system-environment-variable file.

[cmorfin · Accepted solution]

Hi @darcudarcu 

Could you please do the following:

1) restore the original predictive-standalone.yml, so that there is no more the modification you did.

2) edit system-environment-variable.properties file and add the following lines:

working.directory= /C:/INDEFF/WORKFILES/MMAR/thingpredictor

analysis-manager-uri=<Your ThingWorx URI Here>
analysis-manager-appkey=<YOUR Appkey Here>
predictive.http.port=9090
predictive.admin.port=9091
microservices.server=localhost

Note 1:  for analysis-manager-uri you need to have the url to the web socket endpoint, that is for example

ws://123.123.123.45:80/Thingworx/WS

The /WS is important (and missing in your current yml file)

Note 2: you could start by just modifying your current yml file and add the WS at the end of uri.

If it does not work, do the above. It is much cleaner to only update the system-environment-variable.properties file and easier too.

Hope this helps

Christophe

[darcudarcu]

Hi @cmorfin ,

I deleted all the previous folder containing predictor stuff and I followed step 1 and 2.

It worked and I published the models successfully.

However, trying to test the models I encountered two further problems.

1) I am not allowed to fill-in values in Input Data Row Values.
2) I can fill Input Data Row Values but Analysis Job fails.

ApplicationLog:
Job was not found in runningJobs, could cause processing to stop
No results to map
Job not found for update: 601a7b8a-7331-4d45-8e68-ae97d6c063c5 Dynamic deployment completion event is not dispatched

AnalysisJob-->Results-->ImportantFields or predictiveScore--> error message: No parent row exists for the selected structure, loading the closest available parent.

Should I open a new thread for this and accept as valid answer the previous one?

[cmorfin]

Hi @darcudarcu 

Yes I think it is best to open a new thread so others can find the resolution quickly based on the description of this thread.

For the new thread, it can be useful if you add some screenshots of the Analysis model and event configuration (including input and result mapping).

Also can you clarify the 2 points:

1) I am not allowed to fill-in values in Input Data Row Values.
2) I can fill Input Data Row Values but Analysis Job fails.

because in step 1 you wrote you can't fill in value, but in step 2, you wrote you can ... so I am a bit confused.
Also some screenshot can be useful to understand why you can't fill in value.

Thanks

Christophe