first thk u reply.

maybe u tell me the procedure i will test it by myself.

my test size=sum(rows),,,,one query=10,two query(new keyword)=20





2007/4/27, James liu <[EMAIL PROTECTED]>:

These data is true ? hitration is 0??

size=1?

how do u test it?

i think these data is impossible.



2007/4/27, Koji Sekiguchi < [EMAIL PROTECTED]>:
>
> > if u try it. and u find it always 0
>
> Are you talking about evistions?
> If you are talking about evictions and it is always zero,
> it shows that you have enough room in your cache in comparison to your
> load.
>
> By way of example, when we did load test for "size==1 cache", we could
> see:
>
> - queryResultCache-
> lookups: 78626
> hits: 72
> hitratio: 0.00
> inserts: 78554
> evictioins: 1042748
> size: 1
>
> Hope this helps,
>
> Koji
>
>
> James liu wrote:
> > if u try it. and u find it always 0
> >
> > 2007/4/26, Koji Sekiguchi <[EMAIL PROTECTED]>:
> >>
> >> I think cumulative_ counters are the total count of successive
> >> SolrIndexSearchers' cache.
> >> For example, you have a registered searcher and you see:
> >>
> >> inserts: 1
> >> cumulative_inserts: 1
> >>
> >> Then you update your index and do commit, you got a new searcher.
> >> At this moment, non cumulative_ counters come to zero.
> >> Then you search something and you could see:
> >>
> >> inserts: 1
> >> cumulative_inserts: 2
> >>
> >> And if you have enough room for inserting new entries, evictions
> >> might be zero. To see none zero evictions, you can set
> >> a small size for queryResultCache and try to issue many unique
> queries.
> >> For example, set the size to 1, then issue 2 unique queries, then
> >> you can see evictions: 1.
> >>
> >> regards,
> >>
> >> Koji
> >>
> >>
> >> James liu wrote:
> >> > now i use admin gui to moniter cache config
> >> >
> >> > i don't know the difference betwin cumulative_ and no *cumulative_*
> >> > ,,,like
> >> > cumulative_inserts and inserts
> >> >
> >> > and i find evictions always show zero....i m curious when it will
> >> > change and
> >> > its meaning
> >> >
> >>
> >>
> >
> >
>
>


--
regards
jl




--
regards
jl

Reply via email to