Closed MollardMichael closed 3 years ago
Wow
According to Gotenberg's documentation https://thecodingmachine.github.io/gotenberg/#scalability
For the HTML, URL and Markdown endpoints, the API does only 6 conversions in parallel. Indeed, Google Chrome misbehaves if there are too many concurrent conversions. The more concurrent requests, the more 504 HTTP codes the API will return.
I guess issue is related
Can you show, how do you batch requests? And can you try to decrease batch size to 5, for example?
I'm using fp-ts sequences to run then in parallel pipe.
This issue only happen on my kubernetes network. I'm looking more into it. I'll tell you if it comes from the lib or some configuration of axios If I find something.
By the way thanks for the rate limiting. I hadn't seen it
I think you got me wrong :)
I'm not taking any part in Gotenberg's development, this library is a client, which I wrote originally for myself, because there are only two official clients — for Go and PHP.
There is no any rate limiting in this client (maybe I should add it though), this quote is from Gotenberg's documentation.
And I didn't quite understand your mentioning axios... Do you use your own HTTP client, based on axios, with this library? This is possible, but not documented, so I would be quite surprised :) Internally it uses regular Node.js http.request and https.request directly.
I understood correctly the first time. Indeed I made a mistake. I went through your code and I know how to do it now but i'll keep the regular Node.js client.
After some investigation, It turns out that it was a bug in Istio (A service Mesh for Kubernetes).
I'll close this issue. And thank you for the lib. It works great !
I'm trying this library to call Gotenberg and it works great.
I'm generating around 300 pdf and this trying to call Gotenberg in batch of 10s.
When I do this, the calls start to fail with {"errno":-32,"code":"EPIPE","syscall":"write"}.
Do you have any idea why it might happen ?