[Zebralist] zebraidx stop after 6K records - cannot even find error messages anywhere on internet - RESOLUTION

Petrus B. van Bork pvanbork at outerline.com
Tue Mar 17 19:54:33 CET 2009


Dear Fellow List Members:

We found the problem.  The issue was a badly formed MARC record that got 
by all the filters and routines it had passed through.  When we found it 
and deleted it, the whole rest of the MARC batch ran through.

Best,

Petrus


Dear Fellow list Members:

We compiled and run 64 bit Zebra on RHEL5.2.  All well and good.  We
also run the Koha opensource ILS (v 3.00.00) which utilizes MySQL
underneath.  When we go to index our newly imported MARC records all
goes normally until we get:

zebraidx(17758) [log] MARC: Bad directory
zebraidx(17758) [warn] MARC: Base address does not follow directory
zebraidx(17758) [warn] MARC: Bad offsets in data. Skipping rest
zebraidx(17758) [warn] Record didn't contain match fields in  (bib1,
Local-number)
zebraidx(17758) [log] error grs.marcxml.record /tmp .....

...then it goes on and finishes off stating it has inserted 6059 records
(out of about 35.5K) - these are searchable, normally via Koha, btw.
Sometimes, you can get a different # of records, once we inserted 6134
(all from the same, identical initial data).  We thought it might be a
space issue but we have about 4.4G in every directory we can think might
be relevant.  Worse, as previously stated we cannot even find these
error messages documented!

As you can imagine any help would be MOST appreciated.

Regards,

Petrus

P.S. - nothing in our Koha or MySQL logs, either.




More information about the Zebralist mailing list