Going live on Platform.sh is a simple two or three step process.
You can either use the Platform.sh UI or the CLI to configure your project for production. Once you have gone through it once the whole process usually takes a couple of minutes.
The order of operations is not really important, but if you are migrating a site from an existing provider, you should first configure the domain on the Platform.sh side, and only then switch DNS over.
If you are on a Development plan, you cannot add a domain. You will need to upgrade your subscription to a production plan.
Go to your account, click on the small wheel next to you project's name and click on edit.
Choose the plan you want, it will show you the monthly price you will be paying.
You can find more information on pricing on the pricing page.
This is a required step, it will tell the Platform.sh edge layer where to route requests for your web site. You can do this through the CLI with
platform domain:add example.com or using the UI.
You can add multiple domains to point to your project. Each domain can have its own custom SSL certificate, or use the default one provided.
note After you have added your domain, your Master environment will no longer be accessible at
<environment>-<project>.<region>.platform.sh. That's why you should write it down first.
If you require access to the site before the domain name becomes active you can create a
hosts file entry on your computer and point it to the IP address that resolves when you access your master project branch.
To get the IP you can run
ping <environment>-<project>.<region>.platform.sh (with the DNS name you noted in step 2 of the pre-launch check list). In OS X and Linux this will usually be the hosts file in
/etc/hosts in Windows
c:\Windows\System32\Drivers\etc\hosts. You will need to be a admin user to be able to change that file. So in OS X you will usually run something like
sudo vi /etc/hosts. After adding the line the file will look something like:
note Do not put the IP address you see here, but the one you got from the ping command. Also, remember to remove this entry after you have configured DNS!
Sometimes it can take Let's Encrypt a couple of minutes to provision the certificate the first time. This is normal, and only means the first deploy after enabling a domain may take longer than usual.
Configure your DNS provider to point your domain to your Platform.sh Master environment domain name.
The way to do so will vary somewhat depending on your registrar, but nearly all registrars should allow you to set a CNAME. Some will call it an Alias or similar alternate name, but either way the intent is to say "this domain should always resolve to... this other domain". Add a CNAME record from your desired domain (
www.example.com) to the master environment hostname you wrote down earlier.
If you have multiple domains you want to be served by the same application you will need to add a CNAME record for each of them.
Note that depending on your registrar and the TTL you set, it could take anywhere from 15 minutes to 72 hours for the DNS change to fully propagate across the Internet.
The DNS RFC (RFC1033) requires the "zone apex" (sometimes called the "root domain" or "naked domain") to be an "A Record," not a CNAME. But many DNS providers have found a way around this limitation. If you want your site to be accessible with https://example.com and not only https://www.example.com you must use a DNS provider that knows how to do that. Examples of such workaround records include:
These ALIAS/CNAME/ANAME records resolves on request the IP address of the destination record and serves it as if it would be the IP address for the apex domain requested. If the IP address for the destination changes, the IP address for the mapped domain changes automatically as well.
Platform.sh recommends ensuring that your DNS Provider supports dynamic apex domains before registering your domain name with them. If you are using a DNS Provider that does not support dynamic apex domains then you will be unable to use
example.com with Platform.sh, and will need to use only
www.example.com (or similar) instead.
If you are willing to make the
www. version of your site the canonical version (which is recommended), there are various free services that provide blind redirects from
www.example.com domains for any requests sent to their IP. Most of these services are free, and allow you to use a CNAME record to Platform.sh for
www.example.com and an A record to their service at
example.com, which will in turn send a redirect. Examples of such services include:
If you absolutely cannot use a DNS provider that supports aliases or a redirection service, it is possible to use A records with Platform.sh. They will result in a sub-optimal experience, however.
This process has a few limitations:
- Should we ever need to change one of those IPs your configuration will need to be manually updated. Until it is some requests will be lost.
- Directly pointing at the edge routers bypasses their load-balancing functionality. Should one of them go offline for maintenance (as happens periodically for upgrades) approximately 1/3 of requests to your site will go to the offline router and be lost, making the site appear offline.
For that reason using A records is strongly discouraged and should only be used as a last resort.
See the Public IP list for the 3 Inbound addresses for your region. In your DNS provider, configure 3 separate A records for your domain, one for each of those IP addresses. Incoming requests will then pick one of those IPs at random to use for that request.
While not required, it's strongly recommended that you set up health notifications to advise you if your site is experiencing issues such as running low on disk space. Notifications can be sent via email, Slack, or PagerDuty.
It's strongly recommended that you set up automatic snapshots and automatic certificate renewal cron tasks. You will first need to set up an API token and install the CLI as part of the build hook. Then you can easily configure the appropriate cron tasks. The following snippet is generally sufficient but see the the links above for more details, and please modify the cron schedules listed to match your use case.
crons: snapshot: # Take a snapshot automatically every night at 3 am (UTC). spec: '0 3 * * *' cmd: | if [ "$PLATFORM_BRANCH" = master ]; then platform snapshot:create --yes --no-wait fi renewcert: # Force a redeploy at 9 am (UTC) on the 14th of every month. spec: '0 9 14 * *' cmd: | if [ "$PLATFORM_BRANCH" = master ]; then platform redeploy --yes --no-wait fi