Xu Han@AutoMQ
Xu Han@AutoMQ
Solution: - Accelerate LogCache#get - Concurrent read support for LogCache
> this cause one of broker OOM when 1gb/s topic retention time change from 3 days to 12h. after the delete change from 200TB to 18TB. the partition number is...
Replace the quickstart with minio
This PR did indeed greatly improve the performance of WAL in many competitive scenarios. However, it also makes the correctness of WAL in a multi-concurrent context difficult for humans to...
Could you dump the heap?
Purpose of rate limiting: Smoothing out the uploaded traffic to avoid CPU spikes caused by concentrated uploads, resulting in fluctuating transmission delays. The current reasons affecting the speed of partition...
> 3.4 Preload streamSetObject when read from end - When the read is triggered by upper level, then the endOffset is a specific offset. - When the read is triggered...
Will we have a special cache for the node itself? Then we won't need load the SSO which was uploaded by itself.
> 1. endOffset == -1 || endOffset == streamMetadata.getEndOffset 1. If read from the active segment, then the endOffset usually equals streamMetadata.getEndOffset. Because in Kafka, read only pass startOffset and...
cc @SCNieh