lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Furkan KAMACI <furkankam...@gmail.com>
Subject Re: How to run a verification process at pre-commit documents and then commit them into live indexes if they are valid?
Date Fri, 02 Aug 2013 11:18:59 GMT
Thanks for the feedback. I don't want them to be visible at live index
until I remove them. I should make some verification related to Solr. i.

2013/8/2 Mikhail Khludnev <mkhludnev@griddynamics.com>

> I don't think you get it for free in Solr (especially in SolrCloud).
>
> You can experiment with own handler which does Lucene NRT
>
> http://blog.mikemccandless.com/2011/06/lucenes-near-real-time-search-is-fast.htmlit
> can open reader w/o commit that causes flushing rambuffer on disk.
>
> Other approach, is establish something like multi versioned view of the
> index, or scrollable commit history. (I've thought there is a jira, but
> can;t find it). i.e. after commit() and opening new searcher, we can not
> disclosure it and use it for verification internally, and keep older
> searcher accessible for public searches.
>
> One more idea, now prepareCommit() is extracted from commit(), and the
> former one does almost everything, it might be a chance to see (open
> searcher on) files which was flushed by prepareCommit().
>
> All these speculation aren't really helpful, I suppose. Sorry, it's really
> a dev@ question.
>
> However, if the verification procedure is automated, can't you just submit
> hundred of deleteQueries with all undesired spam words before commit?
>
>
> On Fri, Aug 2, 2013 at 12:48 PM, Furkan KAMACI <furkankamaci@gmail.com
> >wrote:
>
> > I use Solr 4.2.1 as SolrCloud. My live indexes will be search by huge
> > amounts of users and I don't want to have anything wrong. I have some
> > criteria for my indexes. i.e. there mustn't be spam documents at my index
> > (I have a spam detector tool), some documents should be at first result
> > page (or within first and second result page, doesn't important) when I
> > search for certain keywords (I don't want to organize them with query
> > elevation, it doesn't important to be at first row or tenth.
> >
> > So I want to run a index verification tool for my indexes. How can I do
> > that with SolrCloud? Should I build a pseudo SolrCloud, make tests on
> them
> > and if everything OK then send to live SolrCloud? Or can I send documents
> > to SolrCloud, don't commit them into live indexes then run that
> > verification tool for not yet committed documents and then if everything
> is
> > OK commit them.
> >
> > What do you prefer me for such cases who wants to run a verification
> > process at their indexes before commit them into live indexes?
> >
>
>
>
> --
> Sincerely yours
> Mikhail Khludnev
> Principal Engineer,
> Grid Dynamics
>
> <http://www.griddynamics.com>
>  <mkhludnev@griddynamics.com>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message