Alfredo, I've been testing this, It isn't working for me -- did you do anything special when indexing as well? On Thu, Feb 9, 2017 at 5:47 AM, Alfredo Sanjuán <alfre@ibdinternet.com> wrote:

Just to let you know that I solved this problem incrementing the max record size allowed in zebrasrv. See -k option: 1024KB by default as configured in Debian packages, too low if you have records with a lot of items.

Just in case someone has the same problem...

Thanks!

El 08/02/2017 a las 12:29, Alfredo Sanjuán escribió:

...

I have a big biblio record (more than 9999 bytes, a lot of items attached) that zebra refuses to index.

I got this error when searching:

Use of uninitialized value $self in pattern match (m//) at /usr/share/perl5/MARC/File/XML.pm line 441. ERROR DECODING RECORD - Empty String at /usr/share/perl5/MARC/File/XML.pm line 448. : at /usr/share/koha/lib/C4/Search.pm line 1920.

and the biblio record doesn't show on the results.

I rebuild the index with: rebuild_zebra.pl -b -x -r -v

Is this a known problem? I'm running Koha 16.11.03.000.

-- Alfredo Sanjuán - IBD Internet Tel: 943376085 _______________________________________________ Koha mailing list http://koha-community.org Koha@lists.katipo.co.nz https://lists.katipo.co.nz/mailman/listinfo/koha

On Tue, Apr 11, 2017 at 5:07 AM, Alfredo Sanjuán <alfre@ibdinternet.com> wrote:

Hi Barton.

No, nothing more than incrementing the maximun record size of zebrasrv. In my case, just by trial and error "-k 104857600" is the minimun size to allow for about 2900 items in a problematic record.

Oh. The documentation says that -k is in kilobytes, which would put your number at just under 105 GB ;-)... if that doesn't take care of a 2900 item record nothing should... But then, documentation *has* been known to be wrong. I'll give it a shot on my test system. --Barton

As a side note, the Z39.50 server is broken too because of this same problem about very big zebra record sizes.

Yeah... that should definitely be a syspref or something...