Our machines have around 8gb of ram and our index is 25gb. What are some
good values for those cache settings. Looks like we have the defaults in
place...
size="16384"
initialSize="4096"
autowarmCount="1024"
You are correct, I am just removing the health-check file and our
loadbalancer prevents any traffic from reaching those nodes while they
are replicating.
On 12/8/10 4:41 PM, Chris Hostetter wrote:
: What am I doing that Solr already provides?
the one thing i haven't seen mentioned anywhere in this thread is what you
have the "autoWarmCount" value set to on all of the various solr internal
caches (as seen in your solrconfig.xml)
if that's set, you don't need to manually feed solr any special queries,
it will warm them automaticly when a newSearcher is opened.
this assumes of course that the SolrCore has old caches to warm forom --
ie: if you use normal replication with an existing SolrCore.
You've made refrences to taking slaves out of clusters using before/after
hooks of your own creation -- as long as this is just stoping traffic from
reaching the slave then auto warning should work fine for you -- if you
are actually shutingdown the SolrCore and starting up a new one, then it
wont -- and you are probably making extra work for yourself.
-Hoss
