hangc0276 opened a new pull request #2963:
URL: https://github.com/apache/bookkeeper/pull/2963


   ### Motivation
   When a bookie restart, the garbageCollectorThread will rebuild 
entryMetadataMap from all the entry log files in ledger directory. For normal 
case, it will extract the EntryLogMetadata from the index in entry log file. 
However, if there's no index, then fallback to scanning the entry log file. 
   
   In user's production environment, the log files without index occupied 4%. 
The total entry log files is 80000, and the log files without index is 3000. 
The default entry log file size is 2GB, and the garbageCollectorThread will 
read 3000 * 2GB = 6TB data without speed limit, which will cause ledger disk IO 
util runs high for dozens of minutes and affect ledger read and write latency.
   
   ### Modification
   1. Add read speed rate limiter for scanning entry log file in 
entryMetadataMap rebuild.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to