[ckan-dev] sqlalchemy migrate, dependency issues.

Thu Jan 27 19:44:55 UTC 2011

Thanks to Friedrich's tips I'm making good progress on this in the
code and hopefully it should be quite smooth. I'll progress this with
David Raznick tomorrow when I see him.

http://ckan.org/ticket/932

David

On 27 January 2011 16:48, William Waites <ww at styx.org> wrote:
> * [2011-01-27 09:11:22 +0000] Seb Bacon <seb.bacon at gmail.com> écrit:
>
> ] > The issue I see with 1 is that if the application is started as an apache
> ] > wsgi script with multiple processes there could be a lot of pain (we do not
> ] > want to upgrade a database from 2 processes) and this would be hard to
> ] > debug.  We would need some kind of update db lock.
> ]
> ] For (3), the update should indeed happen as part of our standard
> ] deploy process, which means it should be scripted.  Obviously a failed
> ] migration should cause a rollback to the previous version.
>
> More complicated, but safer, is the strategy that we take when
> upgrading HMG. There are two databases for each site and two copies
> of the code base. An upgrade process looks like:
>
>        - webapp is running out of db1, code1
>        - make db1 read only (need a better way to do this, mostly done
>          with the apache config restricting POST/PUT/DELETE)
>        - dump db1, restore into db2
>        - upgrade code2
>        - upgrade db2
>        - restart web app out of code2 pointing at db2
>        - re-enable write operations
>
> Everything short of the last step is easy to back out of.
>
> -w
> --
> William Waites                <mailto:ww at styx.org>
> http://eris.okfn.org/ww/         <sip:ww at styx.org>
> F4B3 39BF E775 CF42 0BAB  3DF0 BE40 A6DF B06F FD45
>
> _______________________________________________
> ckan-dev mailing list
> ckan-dev at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/ckan-dev
>