Having to rebuild Lucene index on production machine

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Having to rebuild Lucene index on production machine

Sam Hokin-3
I've found that sometimes (not always), after copying a database dump from my build machine to my production machine, swapping DBs
and firing up the webapp, that search is broken - nothing is found. I have to rebuild the Lucene index (create-search-index
post-process). Does this makes sense? Has anyone had this problem?

I will note in the current situation I've got postgres 9.5.10 on the production machine and 9.6.8 on the build machine. But I don't
think that's what's going on, and I'm using gzipped SQL dumps to transfer (9.6 -Fc dumps won't load on 9.5). The database loads fine.

Not a massive deal, but it takes quite a long time to rebuild the Lucene index for LegumeMine on the production machine. :)
_______________________________________________
dev mailing list
[hidden email]
https://lists.intermine.org/mailman/listinfo/dev
Reply | Threaded
Open this post in threaded view
|

Re: Having to rebuild Lucene index on production machine

joe carlson
Hi Sam,

For what's it's worth, I've sometimes seen problems with the permission
on the lucene file stored in postgres. There were some issues on 8.?
servers not setting or respecting ownership of large objects. And if
you're using the -O flag on the dump, ownership might not be right.
There may there are still some things I'm not clued in on.

Can you still see a couple of large objects in the db when you do a
\lo_list? If so, try as "grant select on large object <number> to
public;" and see if that frees it up.

Joe

On 03/30/2018 08:55 AM, Sam Hokin wrote:

> I've found that sometimes (not always), after copying a database dump
> from my build machine to my production machine, swapping DBs and
> firing up the webapp, that search is broken - nothing is found. I have
> to rebuild the Lucene index (create-search-index post-process). Does
> this makes sense? Has anyone had this problem?
>
> I will note in the current situation I've got postgres 9.5.10 on the
> production machine and 9.6.8 on the build machine. But I don't think
> that's what's going on, and I'm using gzipped SQL dumps to transfer
> (9.6 -Fc dumps won't load on 9.5). The database loads fine.
>
> Not a massive deal, but it takes quite a long time to rebuild the
> Lucene index for LegumeMine on the production machine. :)
> _______________________________________________
> dev mailing list
> [hidden email]
> https://lists.intermine.org/mailman/listinfo/dev

_______________________________________________
dev mailing list
[hidden email]
https://lists.intermine.org/mailman/listinfo/dev