Automated retries

[gotdan]

One thing that would be useful for API retrieval use cases is a parameter to request a number of automated retries by the plugin when it receives non-auth errors, ideally with a throttling pause before retries in response to 429 errors.

[yan-hic]

Any traction on this ?
Ideally the retriable status list should be configurable i.e. passed within the function (e.g retry_on_status=[429, 503]). I would not hardcode these as some servers may not consider some errors as transient. As the OP, # of retries and timeout would be helpful too.

Workaround: revert to Python function to call http endpoints.

[adriens]

Hi, I also faced the 429 issue today.

As a workaround, would it be possible to apply a delay as part of the http call so we could avoid the 429 ?

[lmangani]

Thanks for the reports folks! We definitely need to add some type of retry strategy as an option. Contributions are welcome!

[yan-hic]

Sorry but dunno c++. I reverted to a python function for the same (https://duckdb.org/docs/stable/clients/python/function.html)

[lmangani]

@yan-hic no problem we'll add this as soon as time allows! Thanks again for the useful report 👍