[Koha-zebra] Re: Zebra Sorting Problems

[Sebastian Hammer]

Hi Joshua,

I've never seen an update run just terminate because of a missing match 
key, but then I probably never have provided a record without one. It 
does seem a bit draconian to just terminate.

If you are just aiming to demonstrate search/sort, just remove the 
recordId line in zebra.cfg and start over!

But it'd probably be fun to know why you have a record withoyt an ID 
number in there.

--Seb

Joshua Ferraro wrote:

> On Tue, Mar 21, 2006 at 12:33:03AM -0500, Sebastian Hammer wrote:
>  
>
> > Downloading.....
> >    
> Any luck Sebastian? 
>
> Now I'm trying to index using zebraidx on the cmd line and Tumer Garip's
> zebra.cfg and record.abs files but I'm running into some errors. It
> starts indexing fine but then croaks after about 30 seconds or so:
>
> # zebraidx -g iso2709 -d kohademo update records -n
> <snip>
> 19:16:57-21/03 zebraidx(29467) [log] add grs.marcxml.record records/npl.iso2709 
> 894119
> 19:16:57-21/03 zebraidx(29467) [warn] Record didn't contain match fields in 
> (bib1,Identifier-standard)
> 19:16:57-21/03 zebraidx(29467) [warn] Bad match criteria
> 19:16:57-21/03 zebraidx(29467) [log] zebra_end_trans
> 19:16:57-21/03 zebraidx(29467) [log] sorting section 1
> 19:16:57-21/03 zebraidx(29467) [log] Iterations . . .  94296
> 19:16:57-21/03 zebraidx(29467) [log] Distinct words .  28974
> 19:16:57-21/03 zebraidx(29467) [log] Updates. . . . .     18
> 19:16:57-21/03 zebraidx(29467) [log] Deletions. . . .      0
> 19:16:57-21/03 z
>
> ebraidx(29467) [log] Insertions . . .  28956
> 19:16:57-21/03 zebraidx(29467) [log][app2] zebra_register_close p=0x80b40b0
> 19:16:58-21/03 zebraidx(29467) [log] Records:    1283 i/u/d 1283/0/0
> 19:16:58-21/03 zebraidx(29467) [log] user/system: 1006/28
> 19:16:58-21/03 zebraidx(29467) [log][app2] zebra_stop
> 19:16:58-21/03 zebraidx(29467) [log] zebraidx times: 10.95 10.06  0.28
>
> I'm guessing it dies because it found a record that didn't have whatever
> 'Identifier-standard' is mapped to or something -- any ideas? Is there
> a way to have it just 'skip' a malformed record and proceed?
>
> In the meantime, my other server's about 35 hours into importing a
> dataset of 150K records (it's almost half-way done) using perl-zoom.
> Needless to say, that's not going to ever be an option for initial
> import in the real world -- especially if the index crashes often. Even 
> mysql's faster :-).
>
> I'm also getting some errors occasionally on that import:
>
> no mapping found at position 11 in Mazur, Mont, g0=ASCII_DEFAULT 
> g1=EXTENDED_LATIN at /usr/local/share/perl/5.8.4/MARC/Charset.pm line 134.
> no mapping found at position 53 in Ashton Kutcher, Brittany Murphy, Christian 
> Kane, Mont Mazur, Raymond J. Barry, George Gaynes. g0=ASCII_DEFAULT 
> g1=EXTENDED_LA
> TIN at /usr/local/share/perl/5.8.4/MARC/Charset.pm line 134.
> no mapping found at position 11 in Mazur, Mont, g0=ASCII_DEFAULT 
> g1=EXTENDED_LATIN at /usr/local/share/perl/5.8.4/MARC/Charset.pm line 134.
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
> Oops!  ES: immediate execution failed
>
> Now the Charset error must be related to data that's not yet been defined
> in MARC::Charset, but what's the ES: error about?
>
> Cheers,
>
>  

--
Sebastian Hammer, Index Data
address@hidden   www.indexdata.com
Ph: (603) 209-6853