ok, thanks, so I don't use the -pr option anymore (I used it on the first node). It seems to take some time. running for 150minutes to complete 7% ...
Cheers, Michael On 14.09.2017 15:40, Alexander Dejanovski wrote: > There should be no migration needed, but if you have a lot of data, > anticompaction could take a while the first time. The only way to make that > fast would be to mark all sstables as repaired, and then start running > incremental repair every day or so to have small datasets to anticompact. > But all that data wouldn't really be repaired and no subsequent repair > would actually repair it, unless you run a full repair. > > Do note that full repair performs anticompaction too, and only subrange > repair will skip that phase. > > You should never use "-pr" with incremental repair as it won't mark all > sstables as repaired. It's unnecessary anyway since incremental repair will > skip already repaired tokens, which defeats the benefits of using "-pr". > Just run "nodetool repair" on one node, wait for it to finish (check the > logs for repair completion if nodetool loses the connection at some point). > Only when it is fully finished on that node, move on to the next one. > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@cassandra.apache.org For additional commands, e-mail: dev-h...@cassandra.apache.org
