On Sun, 14 Jan 2018 14:19:54 -0800
Kevin Fenzi <kevin@scrye.com> wrote:

> On 01/11/2018 07:44 AM, Kamil Paral wrote:
> > On Wed, Nov 1, 2017 at 7:48 PM, Kevin Fenzi <kevin@scrye.com
> > <mailto:kevin@scrye.com>> wrote:
> >
> >     On 10/24/2017 03:36 AM, Kamil Paral wrote:
> >     > On Sun, Oct 8, 2017 at 12:14 AM, Kevin Fenzi <kevin@scrye.com
> >     > <mailto:kevin@scrye.com>> wrote:
> >     >> They should be there starting tomorrow.
> >     >>
> >     >
> >     >
> >     > I still don't see any dumps in there. Is it still being
> >     > worked on?
> >
> >     I seem to have made a typo in the script that does this. ;(
> >
> >     Sent in a freeze break request with a fix
> >
> >
> > I and Josef looked at it today, and it seems it contains some very
> > old data - from Jan 2017 to April 2017. We failed to find the
> > script that generates the dump in infra ansible, so we can't really
> > debug what's wrong. Can you please look at it or give us some
> > pointers?
>
> Odd.
>
> It's
> https://infrastructure.fedoraproject.org/cgit/ansible.git/tree/roles/batcave/files/public-db-copy.sh?id=778dae3
> which runs on batcave01 and runs:
> https://infrastructure.fedoraproject.org/cgit/ansible.git/plain/scripts/public-db-copy?id=778dae3
>
> scp db-qa01.qa.fedoraproject.org:/resultsdb-$(date +%F).dump.xz
> /srv/web/infra/db-dumps/resultsdb.dump.xz
>
> and on db-qa01.qa:/backups:
>
> -rw-r--r--. 1 postgres postgres 142589024 Apr 12  2016
> resultsdb-2016-04-12.dump.xz
> -rw-r--r--. 1 postgres postgres  50184088 Jun 20  2017
> resultsdb-2017-06-20.dump.xz
> -rw-r--r--. 1 postgres postgres  50367072 Jan 13 01:59
> resultsdb-2018-01-13.dump.xz
> -rw-r--r--. 1 postgres postgres  50367072 Jan 14 01:12
> resultsdb-2018-01-14.dump.xz
>
> It claims to be the dump from today... so not sure whats going on.

The database for production resultsdb is on db-qa02.qa instead of
db-qa01.qa.

From my read of those scripts, I suspect that is the cause of the
public dumps being old.