Recieving 502 on POST calls

jaredGibb

I am making calls from parabola to xano. i am iterating over a list and subsequently making 1 call per item in my list to update objects, keeping my DB in sync with bubble.

When i make a lot of calls, i occassionally get a 502 error, BUT it's inconsistent. Even on Scale 1X i am getting this. does anyone else get this error when making a lot of POST's to an endpoint.

I have had a ticket in with support for a while on this topic. Hopiing someone in the community has some input here.

Marcel de Pontbriand

Commenting to raise

bcarsley

Yes! a common issue for me as well — commenting to raise !!

Michael Udinski

Hi @Marcel de Pontbriand and @bcarsley — just to clarify you both are also hitting these when using Parabola?

bcarsley

yes absolutely — what's extra confusing is even when I use a rate limiter for the call (and even though the number of records retrieved per min never even gets close to that rate limiter) I still occasionally get that 502…like jaredGibb said it's very inconsistent

Michael Udinski

Thanks for the input @bcarsley. I'll check in with support on Jared's request, hopefully, there's an update there soon. However, I'm curious if there's a route where Parabola can be skipped altogether and allow Xano to handle everything? What type of action are you using in Parabola

Marcel de Pontbriand

@Michael Udinski I am performing a handful of transformation steps, dictionary-type lookups for arrays of references across numerous tables that need to be translated between sets of keys. Also engineering fields from relationship tables to append on to existing records. Finally comparing sets of data between two tables and pending a few triggers( id or no id, modified dates…etc) to post new and patch existing records. This is happening on 6-7 tables. I know I can build in a way to circumenvent ID look-ups in Parabola by setting up a function on the API.

I will say Parabola performs exceptionally well with data transformation, but posting/patching to XANO has been the hiccup. I have not had this error occur when sending to Airtable, Bubble, Webflow, SQL, or any other endpoint.

Michael Udinski

Thanks for the additional details @Marcel de Pontbriand - it sounds like you, @jaredGibb, and @bcarsley are actually all working together on the same project so it's not three different issues :) This would've been good context to know ahead.

I see Jared has an open ticket with support on this and looks like we've been actively working with him to get specific information to troubleshoot better. It sounds like once we get the full information we can begin the process of getting a better look at it. All and all, I'd recommend relying on the support ticket for the most up-to-date information since all the relevant information will be tracked there.

jaredGibb

@Michael Udinski Per my conversation with @Lachlan last night, you all have the pertinent information (that we're able to provide) to resolve the task. If you let us know what else you need from us we'd be happy to provide it.


More details: I have run into this issue before when calling an api endpoint repeatedly from a NODE function. The only wait i was able to overcome the 502 issue was using Axios's Retry along with exponential backoffs.

Michael Udinski

Perfect, thank you! I will let support handle it from here

jaredGibb

As an update here, i was able to reproduce these issues locally on xano by calling 1 endpoint from another endpoint repeatedly. depending on the run, with no other activity on my instance, the 'reciever' endpoint can handle between 950-2000 pings before the first error occurs. Usually this error is a 404 error and looks like the attached photo.

I have tried adding a .01 second pause to the reciever to slow things down to no avail.

context for the test: We occasionally need to update something like 15k-25k records at a time and would do so from either parabola OR potentially xano but it doesnt work from either place atm.

and yes, i also shared this same info with support already. I am just hoping someone in the community may have additional insights here