Hi,
I've built an application which uses Light BPT's to query an API. We run thousands of Light BPT's at once, and have built it alongside the owner of the API. Adjusted the server size so it can handle the amount of requests. Tested in Dev/Test etc and all fine. We pushed to Prod and are now getting a lot of 500's. I spoke to the owner of the API and he said without more info, he doesn't know what specifically could be causing it outside of Production environment being quicker and possibly overwhelming the server that hosts the API? (It's an on prem server for the client).
Any Ideas of suggestions as to what it may be would be greatly appreciated.
Thanks
Hi Aaron,
The first thing come to my mind would be the site property / integration field.
Try to go to the service center and check if the integration field has been set properly and point to the production API url.
Also the API key with credential match the production API url.
Other than that, there's a bit difficult to see what's the cause tho.
Hope this help :)
Credentials etc are all fine - the majority of the time calls do seem to be working. It's just that there are so many calls there are still hundreds of '500' errors.
1ºI would check the id's of the current table, if there are data in prod related to that call.
I can see a case where the bpt before couldn't be executed/aborted or other reason and then there is no data to be retrieved.
2º
api configurations: there might be some limit on the prod env, so the number of calls you are trying to call, surpasses the value registered.
I hope I have helped in some way
Hi @Aaron Gordon
A 500 error code is coming means "internal server error", which could mean that the connection between the Outsystems front end and the company's back end is broken.
so whatever Light BPT's to query an API once you check ,they are giving the correct response or not !!through through postman or have you done the correct implementation of Light BPT's to query an API !!
or very fast trackback the API responses through the service center as @Kay Lun said!
hope this will help you!!
A 500 means the remote server returns a 500. If the "connection is broken", you won't get this response.
First a remark about "We run thousands of Light BPT's at once". Though more Light BPT processes can run at the same time, it's still limited to a reasonable number, like 10 or so (it's configurable, I don't know what the default is). So thousands of processes can be active (i.e. they're elligible to run), but they won't really all run in parallel.
Secondly, a 500 is generated by the remote server. So the owner of the API saying "he doesn't know what [...] could be causing it" is a bit of a cop-out, since they should be able to check their error logs etc. That said, have you tried logging the API calls and check what is actually returned? The 500 error may be accompanied by some payload (sometimes HTML) that describes in more detail what goes wrong.
Agree with Kilian on that the it's the owner.of the API is responsible for fixing the https 500 status error. That is what the error states that that API causes an error.
Correction on the number of concurrent light BPT processes which is 20 per front end server in your environment.
Thanks for the replies so far everyone,
1000's are queued at once and they get executed in parrallel, 20 at a time. The configuration for the actual call isn't an issue as the calls do work the majority of the time.
An issue is that the API software is hosted On prem on the Client's server. It's also prod where I don't have permission to modify logging etc.
I've checked the integrations and the call is taking a loooong time. Maybe it's timing out occasionally.