Launchpad itself

Merge lp://staging/~cjwatson/launchpad/configurable-germinate-base into lp://staging/launchpad

configurable-germinate-base
Merge into devel

Proposed by Colin Watson on 2013-11-07

Status:

Rejected

Rejected by:

Colin Watson on 2014-06-11

Proposed branch:

lp://staging/~cjwatson/launchpad/configurable-germinate-base

Merge into:

lp://staging/launchpad

Diff against target:

195 lines (+29/-22)

4 files modified

configs/development/launchpad-lazr.conf (+1/-0)
lib/lp/archivepublisher/scripts/generate_extra_overrides.py (+9/-8)
lib/lp/archivepublisher/tests/test_generate_extra_overrides.py (+15/-14)
lib/lp/services/config/schema-lazr.conf (+4/-0)

To merge this branch:

bzr merge lp://staging/~cjwatson/launchpad/configurable-germinate-base

High

In Progress

Link a bug report

Reviewer	Date Requested	Status
Colin Watson (community)		Needs Resubmitting on 2014-06-11
William Grant (community)	2013-11-07	Needs Information on 2013-11-20
Review via email: mp+194340@code.staging.launchpad.net

Commit message

Make germinate's seed base location configurable.

Description of the change

== Summary ==

Allow us to configure the seed base URL used by germinate in generate-extra-overrides (by default "http://people.canonical.com/~ubuntu-archive/seeds/") so that we can have it point to http://archive-team.internal/seeds/ instead in production. This will avoid the publisher being reliant on two other machines rather than one for proper functioning, since http://people.canonical.com/~ubuntu-archive/... is now handled by a proxy.

== Tests ==

bin/test -vvct lp.archivepublisher.tests.test_generate_extra_overrides

== Demo and Q/A ==

Run generate-extra-overrides on DF without configuration and verify that it still works. Configure DF to use some different URL with modified seeds instead and verify that the new output of generate-extra-overrides reflects these.

Revision history for this message

Adam Conrad (adconrad) wrote on 2013-11-07:

maintenance-check.py also hits people.canonical.com for the same reason, should probably be fixed to use the same config option.

Testing this should perhaps be done by intentionally firewalling off people.c.c from dogfood, doing a publisher run, and scraping logs to make sure we didn't miss anything. :P

Revision history for this message

Colin Watson (cjwatson) wrote on 2013-11-08:

On Thu, Nov 07, 2013 at 09:25:47PM -0000, Adam Conrad wrote:
> maintenance-check.py also hits people.canonical.com for the same reason, should probably be fixed to use the same config option.

It probably needs an additional one, because it points to
germinate-output as well as seeds and I'm a bit wary of hardcoding their
relative locations.

I'm also wary of changing anything substantial in maintenance-check
until it has tests worth the name, though. I should finish that
refactoring branch ...

Revision history for this message

William Grant (wgrant) wrote on 2013-11-20:

Overriding the config should usually only been done in high-level integration tests. What was the rationale behind having makeSeedStructures() look up the config option, rather than doing it in generateExtraOverrides() or even process()? That would surely make testing less ugly.

review: Needs Information

Revision history for this message

Colin Watson (cjwatson) wrote on 2014-06-11:

I'll redo this on top of lp:ubuntu-archive-publishing, and improve the configuration strategy in the process.

review: Needs Resubmitting