keith-turner opened a new pull request, #5627: URL: https://github.com/apache/accumulo/pull/5627
The manager sends the set of migrations to each metadata tablet server when it scans the metadata table. For a large set of migrations and many metadata tablet servers the cost of repeatedly sending the same data adds up. This change compresses the migrations once and then each rpc to a tserver sends the compressed data. Decompression will happen in parallel on the tservers. The one time cost of compression should eventually be worthwhile because of the reduced amount of data sent from manager to many tservers. A much better way to solve this overall problem is #5416, but that would be a very disruptive to change to make in 2.1. This change and #5626 can improve throughput for this situation in 2.1 w/ minor changes. These changes can probably be discarded when merging to main. -- 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]
