`ChatBot` running in `Poller` mode gets `{:error, "Too Many Requests: retry after 5"}` about telegram

[visciang]

Yes, nothing strange.

FYI in #144 (release 0.22..2) I've removed a incorrect retry logic that was applied also to errors raising from the bot.

BTW If you can hit the problem consistently, please try with 0.22.3 and keep an eye at the logs.

PS: recv_timeout should be strictly greater than the HTTP long polling timeout (get_updates_poll_timeout_s defaults to 30s), for example 40s is fine. I've updated the docs recently.

[visciang]

Hi @samm81,

can you show me a simple bot that triggers the telegram rate limit (I suppose on the "getUpdates" request)?
It will be useful to check the config you are using (for example if you have a custom value for the [:telegram, :get_updates_poll_timeout_s] config), which tesla client etc.

You could start from the telegram/example/example_bot.exs and tweak it to show the issue.
(BOT_TOKEN="...your token..." ./example/example_bot.exs).

Thank you!

[visciang]

Check out #146 .
I've included a retry rule in case of server side throttling.
Released in v0.22.3.

BTW, this should not happen in your case, even with a get_updates_poll_timeout_s = 0 the telegram server does not send me a 429, it just slow me down. Maybe this could happen if you send lots of messages ion a tight loop.

Please investigate a bit more on your side.

I'm closing this issue for now, feel free to ping back if your have any additional details or question.

[samm81]

hi, thanks for the quick response!

definitely nothing crazy for my config, config :tesla, adapter: {Tesla.Adapter.Hackney, [recv_timeout: 30_000]} is all. I'm definitely not sending messages in a tight loop. I did set the max_bot_concurrency: 1, as I was only expecting to have one client (me), maybe that's it?

I will try v0.22.3, seems I was a whole minor version behind (v0.22.0), thanks again!