bardock opened a new issue #1981: Poor autorecovery performance on dedicated hosts URL: https://github.com/apache/incubator-pulsar/issues/1981 We have a cluster of 20 bookies and another cluster of 15 instances running autorecovery. We've just lost a bookie (physically) and right now there are 10k ledgers marked as underreplicated (approximately 100 GB of data). Autorecovery is replicating 50-100 ledgers per minute. Is there any way to improve this performance? There are enough resources (CPU, RAM and network bandwidth), however publish times increased from 10 to 100 ms (pct99). We are using EC2 instances ([i3.2xlarge](https://aws.amazon.com/ec2/instance-types/i3/#Product_Details) for bookies and [c5.large](https://aws.amazon.com/ec2/instance-types/c5/#Product_Details) for autorecovery) and bookkeeper version 4.3.1.91-yahoo with the default config except these settings: ``` dbStorage_readAheadCacheMaxSizeMb=4096, dbStorage_rocksDB_blockCacheSize=4294967296 flushInterval=10000 ```
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services