Transition a site to GOV.UK
When a site is going to move to GOV.UK, there are two ways that the old site can be redirected. They can do it themselves, or they can repoint the domain at us. This page is about the latter.
The Transition app exists to allow editing of mapping of old URLs to pages on GOV.UK. These mappings are stored in a database and used by Bouncer to handle requests to those old domains.
This page covers adding a site so that we can handle traffic to it, or changing the configuration of an existing site in the Transition app.
Changing the configuration of a site that is already in the Transition app
The transition-config README gives more details on how to edit the relevant transition configuration file.
Checklist for transitioning a new site
1) Add a site to the Transition app
Follow the instructions in the transition-config README.
2) Consider AKA domains
If the department want to use the side-by-side browser to preview how their redirects will appear after transitioning they will need to set up AKA domains.
They will need one new CNAME DNS entry for each domain/subdomain they wish to preview according to the following pattern:
- www.domain.gov.uk → aka.domain.gov.uk
- sub.domain.gov.uk → aka-sub.domain.gov.uk
There are lots of examples of these in hosts currently configured in Transition.
These AKA domains should be CNAMEd to: bouncer-cdn.production.govuk.service.gov.uk
Previously we used a redirector-cdn address which reached its capacity and will now fail if added as the
CNAME record for any new domains.
3) Get a list of old URLs
In order for us to redirect anything but the homepage, we need a list of URLs for the old site so that they can be mapped. In the past, this is something GDS did for the organisation as it requires some technical skill and experience.
Getting traffic logs from a transitioning organisation is the best option, as we can get a list of URLs we know are actively used. If we can’t get this then there are other options.
One option is to get a list of the URLs from the Internet Archive. The
archive_lister gem can do this for
you. Sometimes the Internet Archive doesn’t have any data, so try the domain
name with http or https, or with and without the www..
An alternative is to crawl the site with a crawler like Anemone, though for a large site this might take several hours. This will give you a list of URLs for a domain:
$ anemone url-list 'transitioning-site.gov.uk'
Note: This will include 404s, 301s, etc.
4) Clean up URLs
Strip paths and pattern
There are lots of file formats we don’t want to provide mappings for, like static assets, images, or common spammy/malicious crawlers. These can be stripped using the strip_mappings.sh script.
Query parameter analysis
From your set of URLs, you can attempt to identify significant query string
parameter names and then add them to the site configuration file in
transition-config. A query string parameter is considered significant if it
significantly changes the content seen on the old site and/or it would be
mapped to a different new URL.
There are some transition-config scripts to help analyse query param usage:
Some common examples of significant parameters:
- article ID
- attachment ID
- document ID
Some common examples of non-significant parameters:
- pagination
- analytics
- search queries
5) Add the old URLs as mappings
Ideally, any significant query string parameters should be identified and added to the site before adding the mappings. This is because URLs are canonicalised and then deduplicated before being saved and part of canonicalisation is to remove non-significant parameters. However, significant query string parameters can be added later; after they have been imported adding old URLs which include the parameters will be canonicalised using the new list of significant parameters. Note that mappings created under the previous list won’t be removed and in some (complicated) situations won’t be used by Bouncer.
To add or edit mappings, you will need the “GDS Editor” permission in
GOV.UK Signon for the Transition app. This lets you edit any site,
rather than just ones belonging to your organisation. You can then go
the transition app, find
the site and click Add mappings to add them in bulk.
To allow mappings which redirect away from GOV.UK, you’ll need to add the site
into the allowlisted hosts in Transition. You’ll need the admin permission in Transition to be able to see this page.
6) Get the organisation to do the mapping work
By default, the mappings will present an archive page to users visiting the old URL. The objective is to get users to somewhere that best serves the need fulfilled by the old page. Usually this means redirecting them to the page on GOV.UK or elsewhere. It is really important that this is done by people who understand the users and content.
7) Get the organisation to lower the TTL on the DNS records a day ahead
In order to cleanly switch the domain from the old site, the TTL needs to be low enough that there isn’t a significant period where some users will get the old site and some get the new one. This is important for several reasons, including user experience and giving a professional impression to stakeholders. We normally ask for this to be done a day in advance, and to be lowered to 300 seconds (5 minutes). It can be raised again once everyone is happy there is no need to switch back - normally the day after.
8) Point the domain at us
Once the site has been imported successfully, the domain can be pointed
at us by the organisation. For hostnames which can have a CNAME
record, this is bouncer-cdn.production.govuk.service.gov.uk.
Domains at the root of their zone can’t use CNAME records, so must use
an A record and point at one of the Fastly GOV.UK IP
addresses.
If the site is one that was administered by GDS (e.g. theorytest.direct.gov.uk), you will need to update and re-deploy the DNS config.
You’ll need to create a TLS certificate in Fastly for HTTPS domains, otherwise users will see a certificate error when being redirected from an external HTTPS URL to GOV.UK via Bouncer. Read how to request a Fastly TLS certificate
9) Get the organisation to monitor the traffic data in the Transition app
There are two things that need to be responded to:
- high numbers of 404s - this means a mapping is missing
- high numbers hitting 410s - this means the old page is popular and should perhaps be redirected instead
Further reading
The transition checklist covers the whole process of transitioning a site from the technical side. There’s a full version for complex sites and a slightly simplified one - we probably only need the simpler one for upcoming transitions.