Hi folks, I think that the timestamp should be rounded down to a minute (or whatever) to avoid trashing the filter query cache
Cheers, Siegfried Goeschl On 25 May 2014, at 18:19, Steve McKay <st...@b.abbies.us> wrote: > Solr can add the filter for you: > > <requestHandler ...> > <lst name="appends"> > <str name="fq">timestamp:[* TO NOW-30SECOND]</str> > </lst> > </requestHandler> > > Increasing soft commit frequency isn't a bad idea, though. I'd probably do > both. :) > > On May 23, 2014, at 6:51 PM, Michael Tracey <mtra...@biblio.com> wrote: > >> Hey all, >> >> I've got a number of nodes (Solr 4.4 Cloud) that I'm balancing with HaProxy >> for queries. I'm indexing pretty much constantly, and have autoCommit and >> autoSoftCommit on for Near Realtime Searching. All works nicely, except >> that occasionally the auto-commit cycles are far enough off that one node >> will return a document that another node doesn't. I don't want to have to >> add something like this: timestamp:[* TO NOW-30MINUTE] to every query to >> make sure that all the nodes have the record. Ideas? autoSoftCommit more >> often? >> >> <autoCommit> >> <maxDocs>100000</maxDocs> >> <maxTime>7200000</maxTime> >> <openSearcher>false</openSearcher> >> </autoCommit> >> >> <autoSoftCommit> >> <maxTime>30000</maxTime> >> <maxDocs>5000</maxDocs> >> </autoSoftCommit> >> >> Thanks, >> >> M. >
