I made a support ticket one week ago about this but didn’t get any feedback for now.
Here is my problem :
I am using Node Red on an EDGE IO to do product traçability in my shopfloor.
We are using a flashing device (Keyence Camera) to scan Datamatrix. The camera is connected to the EDGE IO with ethernet cable. Then I’m using Node Red to transform the data and send it to a TULIP Table thanks to the TULIP API Node “Create record”.
My issue is that I have often an error response forme my Node Creat Record : “Error : Response status code 500”.
Sometimes, it works fine and sometimes not…
When I look into the API documentation, it says : “The server encountered an unexpected error.”
I don’t know where the problem can come from… But the impact for us is important as we are loosing important traçability data.
I am no Node-RED expert myself, but a couple things that may be helpful to look at /consider here:
500 Tulip API error is not unusual and can occur, during things like maintenance updates. From my understanding, it is impossible to guarantee 100% API availability, so potentially that could explain some of the inconsistency you’re seeing here.
If this is just normal error anomalies, and the tulip-tables node does not offer out-of-the-box error retries, then you would have to build logic in the Node-RED flow to automatically retry ~5 times when receiving a 500 HTTP error, or store this data in a local file somewhere to later retry manually.
@paul.genoud another option, though not ideal since you already have your Keyence camera, but we have a more out of the box integration with Cognex Dataman that can run on Edge IO with their driver- you can see more details here. Just another thought / if you are looking to do more of these integrations in the future!
Nothing jumps out at me as wrong. Usually a 500 error means something is malformed (it doesn’t like it).
I ran into this after an LTS upgrade. Apparently I setup a connector function not quite right, but it worked anyway. After the upgrade it rightly complained but it took me a couple of hours to figure out what the problem was.
I can tell you that we have avoided linked records, which is what I assume the _link_left_column is for a few reasons. One big one was the ability for it to be used in CSV imports which I believe still is not supported (we’re currently on LTS 12.9 but about to upgrade to 13.x).
I’m not sure what the error would be if you tried to create a record with the same ID as a previous record, but as you’re using millisecond timestamps, that seems highly unlikely
Maybe you can collect a few JSON payloads of ones that worked and ones that didn’t and see if you can spot a difference.
