Indeed, I don't understand the caveat too, but I can imagine that is related with some algorithm to trigger a full sync if necessary.
I will waiting for 5.3 to do the upgrade and have this configuration available. —/Yago Riveiro On Sun, Aug 23, 2015 at 3:37 AM, Shawn Heisey <apa...@elyograg.org> wrote: > On 8/22/2015 3:50 PM, Yago Riveiro wrote: >> I'm using java 7u25 oracle version with Solr 4.6.1 >> >> It work well with > 98% of throughput but in some full GC the issue arises. >> A full sync for one shard is more than 50G. >> >> There is any configuration to configurate the number of docs behind leader >> that a replica can be? > It looks like the number of docs is configurable in 5.1 and later: > https://issues.apache.org/jira/browse/SOLR-6359 > There is apparently a caveat related to SolrCloud recovery, which I am > having trouble grasping: > "the 20% newest existing transaction log of the core to be recovered > must be newer than the 20% oldest existing transaction log of the good > core." > Thanks, > Shawn
