Different request limitation on different companies

[David Tandberg-Johansen]

Hi,

We have exprienced for our customers, that importing new order lines, that there is diferent limitations for different company(Our customers). Could someone please explain why there are different limitations, and if we can use the API to check which level the companies have?

As an example:

We do order import for new orders for 2 different companies.

Theese orders are new, so we send in lines as a value of [OrderLine_Insert_Input!]!,

The value of values is a array of OrderLine_Insert_Input!.

For one company we can send only 25 lines, but for another company we can send over 100 lines. When we go over the treshold of lines, we get internal server error.

So why is it different between the companies

[David Tandberg-Johansen]

Hi @Øyvind Årseth ,

Thank you for your reply.

So there are no limits by company in the hole stack, from GraphQL to the Database?

I haven't any specific right now, since we have taken steps to not get a error situation for our customers.

But we are in a process where such situation would be critical of how we set up things from our side.

Could I maybe later send a email to cloud.api.support@visma.com, or create a new thread here in the forum.

[Øyvind Årseth · Accepted solution]

@David Tandberg-Johansen There are no company-specific limits directly, no. There is a rate-limit on requests per second per client session (access token) of 15, with occasional bursts of 25 allowed. You would see a 429 response in these cases. 
There is a size limit on the toal request (including attachments) of 15 MBs, where you would see a 413 in the response. 
In case of mentioned timeouts, you should still not get a 500-error.
Feel free to create a case if it happens again or with any regularity. As much info as possible is useful, customerID, timestamp + what the request is meant to do etc. 

[Øyvind Årseth]

Hi @David Tandberg-Johansen There are no differences between companies per se, so it must be something else.
This is not a quota/rate limiting thing, it would give 429 - Too many requests. Is the described behaviour consistent on the companies you mention? If you can provide a timestamp for an occurence (customerID would also help), we can investigate and see if we can catch what happened. You may experience timeouts if your request exceeds 30 secs, but they would normally give a 504 I think.