how many langfuse.flush() will retry for network faliure ?

[samaiLLM]

We have a self-hosted Langfuse instance. We tested a scenario where our AI application could not reach Langfuse (simulated by updating a WAF rule to block access). During this test, we observed CPU utilization rising, eventually leading to the application crashing. We are concerned that if Langfuse becomes unreachable in production, our AI application could fail. How can we prevent this in production? Additionally, how many retry attempts does langfuse.flush() perform, and what would be the root cause for the AI application requiring a restart in such scenarios?

[jannikmaierhoefer]

Hi @samaiLLM, Langfuse is built to be asynchronous and should not affect your running application if Langfuse is not reachable (except for fetching prompts).

Could you share some more context on what led your application to fail in your case?