BTW. 3.1.6 we've fixed two major bugs in cache mode.
So I affraid that evebody have to drop database and reindex :-(
We sorry about that.
The problem is that first indexing did not put URL timestamps into
/var/raw/del.log
This caused a problem that the same URL distributed again and again
everytime when spitter is running. This means that the same word in the
same URL is found several times.
Fredy Kuenzler wrote:
>
> Dear Alexander and all
>
> I'm currently testing 3.1.6-pre2, indexing a big number of
> domains, using the new cache mode. Currently appx. 500'000
> Documents total, 1/3 already indexed, 2/3 still to go.
>
> After indexing overnight with 4 concurrent indexer processes
> (seems to work fine) and running splitter afterwards, I
> expericend a strange issue. It seems, previous queries are still
> beeing cached and won't be updated.
>
> Here is the example:
> Yesterday afternoon I queried "fussball". 49 Results.
> After indexing overnight and splittering, I queried "fussball"
> again. Still 49 Results. If I query "fussball" and "sport", it
> shows more results: fussball: 997 sport: 12881
>
> I checked with different browsers to remove cache entries and
> also with a previously unused proxy server - no difference.
>
> You might try it out
> http://potato.webtourist.net/cgi-bin/pftest/search.cgi
> to figure out whether I'm expericening a fata morgana or not :-)
>
> Thanks
> Fredy
>
> BTW. There is another minor issue in search.cgi: The 2nd page is
> numbered from 1 to 20 instead 21 to 40 and so on (default
> settings).
--
Alexander Barkov
IZHCOM, Izhevsk
email: [EMAIL PROTECTED] | http://www.izhcom.ru
Phone: +7 (3412) 51-32-11 | Fax: +7 (3412) 51-20-80
ICQ: 7748759
______________
If you want to unsubscribe send "unsubscribe udmsearch"
to [EMAIL PROTECTED]
