Mike,

On first inspection, it looks like rerunning the import chain and filtering on mrd records will be the most straight forward for us. Thanks for considering all of our options. I'll report back about how long it takes to do a repair like this.

-dale

Mike Rylander wrote:
On Fri, Jun 12, 2009 at 10:36 AM, Frances Dean
McNamara<[email protected]> wrote:
I believe NONE of the records loaded got rows.  Perhaps trying the bib_ids.list 
for ranges of records to see if it works.  You say that would be a really long 
thing?  Does that do more thand the metabib.rec_descriptor inserts?


Right.  I gave you the more complicated SQL so you could grab a fresh
set of just the remaining to-be-fixed records if you have to stop the
process or it dies.  It re-ingests the entire record, basically
rewriting it.  This, of course, assumes that the mrd portion of the
ingest process is not broken in some way and that re-ingesting will
fix things. you can test on one record to find out, of course.

Or we could use your first suggestion.  We actually do have the original input 
files broken down to about 59 files I think.  I will talk to Dale when he is 
in.  If we figure out which way to do it we can probably script it.


This will be faster, but more involved.  Pop in here if you run into trouble.

--miker

Thanks.

Frances

-----Original Message-----
From: [email protected] 
[mailto:[email protected]] On Behalf Of Mike 
Rylander
Sent: Friday, June 12, 2009 9:03 AM
To: Evergreen Development Discussion List
Subject: Re: [OPEN-ILS-DEV] No rows in metabib.rec_descriptor after loading 
bibs.

On Fri, Jun 12, 2009 at 9:47 AM, Frances Dean
McNamara<[email protected]> wrote:
Ah, I see, it is pulling out stuff like type of date, date1, date2 from the 
fixed field, bib lvl, enc lvl, looks like mainly 008 type of data?  Is there a 
script we could run for a while to build this?  We have over 5 million bibs.  
Thanks.  Frances

You can use the stock ingest scripts, and avoid intermediate files by
doing something like:

$ marc2bre.pl --marctype=xml [... other options] | direct_ingest.pl |
grep '"__c":"mrd"' > mrd.ingest

for 1.4.  If you're working with 1.2, you'll need to change the grep,
but that will get you the mrd data.

Another other option is to get a list of all the bibs that lack 'mrd's
in a file and re-ingest them.  In psql:

 \t
 \o /tmp/bib_ids.list
 SELECT b.id FROM biblio.record_entry b LEFT JOIN
metabib.rec_descriptor d ON (b.id = d.record) WHERE r.id IS NULL b.id
0 AND NOT b.deleted;
 \o
 \t

Then as the opensrf user, at the command line:

$ for i in `cat /tmp/bib_ids.list`; do echo "request open-ils.ingest
open-ils.ingest.full.biblio.record $i"; done | srfsh

That will take, um, a really long time to complete.  If anything
happens to stop that, just start over with a fresh bib_ids.list file
-- the SQL above will give you only te records that are missing an
'mrd'.

--miker

-----Original Message-----
From: [email protected] 
[mailto:[email protected]] On Behalf Of Frances 
Dean McNamara
Sent: Friday, June 12, 2009 8:37 AM
To: Evergreen Development Discussion List
Subject: Re: [OPEN-ILS-DEV] No rows in metabib.rec_descriptor after loading 
bibs.

Dale has not replied to this yet, but I don't think we have kept the interim 
files due to file space (big database).

Are their scripts needed to regenerate that table from what is in the db now?  
What is in that table anyhow?

Frances McNamara
University of Chicago

-----Original Message-----
From: [email protected] 
[mailto:[email protected]] On Behalf Of Mike 
Rylander
Sent: Friday, June 12, 2009 7:12 AM
To: Evergreen Development Discussion List
Subject: Re: [OPEN-ILS-DEV] No rows in metabib.rec_descriptor after loading 
bibs.

On Thu, Jun 11, 2009 at 7:23 PM, Dale Arntson<[email protected]> wrote:
Hi All,

After loading our bib records into evergreen, I got zero hits on searches in
the evergreen client. I traced the problem back to the fact that there are
no records in the metabib.rec_descriptor table. The other metabib tables
seem fully populated. Here are the flags I used in parallel_pg_loader.

perl parallel_pg_loader.pl -order bre -order mrd -order mfr -order mtfe
 -order mafe -order msfe -order mkfe -order msefe -autoprimary mrd
-autoprimary mfr -autoprimary mtfe -autoprimary mafe -autoprimary msfe
-autoprimary mkfe -autoprimary msefe

Any ideas what I did wrong? Any ideas how to fix it? The bibs took a long
time to load. I would rather not redo it, if I don't have to.
If you have the intermediate bib processing files, particularly the
output of direct_ingest, make sure that file contains 'mrd' rows.  For
OpenSRF 0.9 (Evergreen 1.2.x), those will start with '/*--S mrd--*/',
and in OpenSRF 1.0.x they will start with '[{"__c":"mrd","__p"'.

If you can find that file and it contains 'mrd' lines, then we can
simply regenerate that table's worth of data. Otherwise you'll need to
reprocess the bib records in order to get that data, but you won't
need to reload the entire dataset.

--
Mike Rylander
 | VP, Research and Design
 | Equinox Software, Inc. / The Evergreen Experts
 | phone:  1-877-OPEN-ILS (673-6457)
 | email:  [email protected]
 | web:  http://www.esilibrary.com



--
Mike Rylander
 | VP, Research and Design
 | Equinox Software, Inc. / The Evergreen Experts
 | phone:  1-877-OPEN-ILS (673-6457)
 | email:  [email protected]
 | web:  http://www.esilibrary.com




Reply via email to