[Koha] What is the point of the Undo Import into catalogue feature?

[Paul A]

At 01:28 PM 12/5/2016 +0000, Raymund Delahunty wrote:
>I ran repeated tests for up to 48 hours or more. I think the indexes were 
>partly updated- 7K out of 13K "remained". I have two files to deal with 
>tomorrow (3K, and 30K small MARC records- bibs and items). I am tempted to 
>try the Undo again to see if it works. I am asking our support company 
>PTFS Europe about any issues there might be with the zebra indexer. If it 
>goes wrong again I'll ask for another index rebuild, and give up on the 
>Undo feature.

I would agree with what François wrote re: Zebra. Just adding that the 
normal Koha "incremental" cron job appears to be bullet-proof for 
additions, and is normally totally reliable for "admin/staff" adds, 
deletes, edits, etc. We run it every minute and it's a matter of milliseconds.

However, we have had times where a complete (bibs *and* auths) re-index was 
required, but by the time the cataloguers told me "what they had done" I 
could not reproduce it. So I now have a weekly cron job for a complete 
re-index. Although it looks "long" on an SSH screen, the actual down-time 
of the Zebra server is just a few seconds, so our cataloguers have been 
told to call me if in doubt, and I do it manually. I'm sure PTFS can sort 
this out for you.

Paul


>Ray Delahunty.
>
>
>-----Original Message-----
>From: Koha [mailto:koha-bounces at lists.katipo.co.nz] On Behalf Of Francois 
>Charbonnier
>Sent: 05 December 2016 13:14
>To: koha at lists.katipo.co.nz
>Subject: Re: [Koha] What is the point of the Undo Import into catalogue 
>feature?
>
>Hi,
>
>Have you waited long enough for zebra to re-index your catalogue ? To me, 
>what you are describing is not faulty. Once you reverted an imported 
>batch, it takes time to zebra to remove the records from the indexes.
>Especially if you work with large files.
>
>If you waited long enough and the indexes never got updated, I would say 
>it's faulty, yes. But it you reverted the import and search for records 
>right away, I would say it's just zebra that hasn't been able to reindex 
>everything...
>
>
>François Charbonnier,