@RunLLM if the backlog on topic is around 15 to 20 gb .will the readtimeout for the reader should be more ?

It seems likely that if your Pulsar topic backlog is 15–20 GB, you may need to increase the read timeout for the reader, especially if reading from cold storage or if the broker needs to fetch data from tiered storage or disk. Large backlogs can increase the time required for the broker to serve read requests, so a higher timeout can help avoid premature timeouts and failed reads. However, the optimal timeout depends on your storage performance and network conditions. There is no specific formula in the data sources, but similar concerns are discussed for BookKeeper and Pulsar read timeouts in high-latency scenarios, and Pulsar sets a default read timeout of 30 seconds, which may need adjustment for large backlogs or slow storage systems (BookKeeper PR 4176). Would you like more detail on how to tune these timeouts or best practices for large backlogs? Sources: - BookKeeper PR 4176 👍 0 button 👎 0 button 💡 Hint: Mention @runllm in the thread for followups.