We experienced exact opposite issue on Solr 4.10

Check my comments in https://issues.apache.org/jira/browse/SOLR-9437

I am not sure if issue was fixed in Solr 6

I do be interested in tracking down patch for this.

On Sep 14, 2016 3:04 PM, "Erick Erickson" <erickerick...@gmail.com> wrote:

> Weird indeed. Optimize _shouldn't_ be necessary if the index was
> rebuilt from scratch after changing something like DV, but in a mixed
> set of segments I'm not sure what would happen. Perhaps one of the
> Lucene folks can chime in?
>
> Best,
> Erick
>
> On Wed, Sep 14, 2016 at 9:22 AM, Markus Jelsma
> <markus.jel...@openindex.io> wrote:
> > Well, it could be that indeed. I know i enabled docValues on that field
> three and a half months ago. But usually when i do that, i force an
> optimize.
> >
> > On the other hand, i'd reckon that in the past few months, all segments
> should have been merged with another one at least once because data keeps
> streaming in. But i'm not sure it would anyway.
> >
> > Thanks,
> > Markus
> >
> > -----Original message-----
> >> From:Erick Erickson <erickerick...@gmail.com>
> >> Sent: Wednesday 14th September 2016 17:22
> >> To: solr-user <solr-user@lucene.apache.org>
> >> Subject: Re: Facetting on a field doesn't work, until i optimized the
> index
> >>
> >> That's strange....
> >>
> >> Is there any chance that the schema changed? This is _really_ a shot
> >> in the dark, but perhaps the optimize "normalized" the field
> >> definitions stored with each segment.
> >>
> >> Imagine segments 1-5 have one definition, and segments 6-10 have a
> >> different definition for your field. Optimize would have to resolve
> >> this somehow, perhaps that process made the magic happen?
> >>
> >> NOTE: I'm not conversant with the internals of merge, so this may be
> >> totally bogus......
> >>
> >> Best,
> >> Erick
> >>
> >> On Wed, Sep 14, 2016 at 6:15 AM, Markus Jelsma
> >> <markus.jel...@openindex.io> wrote:
> >> > Hello - we've just spotted the weirdest issue on Solr 6.1.
> >> >
> >> > We have a Solr index full of logs, new items are added every few
> minutes. We also have an application that shows charts based on what's in
> the index, Banana style.
> >> >
> >> > Yesterday we saw facets for a specific field were missing. Today we
> checked it out until we reduced the facet query just to
> facet=true&facet.field=FIELD, but it returned nothing of use, just an empty
> set of facets.
> >> >
> >> > My colleague suggested the crazy idea to optimize the index, i
> protested because it is no use, numDoc always equals maxDoc and the
> optimize button was missing anyway. So i forced an optimize via the URL,
> and it worked, the facets for that field are now back!
> >> >
> >> > Any ideas? Is there a related ticket?
> >> >
> >> > Thanks,
> >> > Markus
> >>
>

Reply via email to