Replicate application data locally for development
Dumps are generated from production data in the early hours each day, and can then be downloaded from integration (AWS). The process is managed by the replicate-data-local.sh script within the govuk-puppet repository.
The Licensify and Signon databases aren’t synced out of production because of security concerns. Mapit’s database is downloaded in the Mapit repo, so won’t be in the backups folder.
If the download process takes longer than an hour it will stop due to your AWS token expiring. When this happens simply restart the process, which will generating a new token, it should resume downloading from where it left off.
Pre-requisites to importing data
To get production data on to your local VM, you’ll need to have either:
- access to integration via aws; or
- database exports from someone that does.
The AWS setup guide covers using the interface in full, however if you only want to do developer replication you will need to:
1. Log into AWS
2. Setup MFA/2FA on your device
Navigate to the users list in IAM and find your username (“IAM > Users” >
If you don’t see an ID next to “Assigned MFA device” click the edit button and set one up.
3. Create an access token
Under “Access keys” click “Create access key”. Your secret is only displayed once, however if you fail to note it down just remove it and create another.
4. Install the AWS CLI application locally
On macOS this can be done using
brew install awscli. This is being installed locally with the assumption you will be doing a two step backup process, i.e. download the files on your local machine and then update you VM. This is the recommended process as the download is quicker to you local machine for most users.
5. Setup your AWS access config
[profile govuk-integration] role_arn = <Role ARN> mfa_serial = <MFA ARN> source_profile = gds region = eu-west-1
role_arnshould be one of the following depending on the level of access you have been configured with here:
The account IDs are in the govuk-aws-data repo’s docs on govuk-aws-accounts
MFA ARN: This is the long string next to “Assigned MFA device” in the AWS console.
6. Create credentials file
[gds] aws_access_key_id = <access key id> aws_secret_access_key = <secret access key>
AWS access key id: This is the ID that we created earlier
AWS secret access key: This is the secret associated with the key we created earlier. If you didn’t note these down you can simply create a new one.
If you have integration access
If you have integration access, you can download and import the latest data by running:
mac$ cd ~/govuk/govuk-puppet/development-vm/replication mac$ ./replicate-data-local.sh -u $USERNAME -F ../ssh_config -n
You may be able to skip the -u and -F flags depending on your setup
The data will download to a folder named with today’s date in
./backups, for example
dev$ cd /var/govuk/govuk-puppet/development-vm/replication dev$ ./replicate-data-local.sh -d path/to/dir -s
You can skip the -d flag if you do this on the say day as the download
Databases will take a long time to download. They’ll also take up a lot of disk space (up to ~30GB uncompressed). The process will also take up a bunch of compute resource as you import the data.
If you don’t have integration access
If you don’t have integration access, ask someone to give you a copy of their
dump. Then, from
dev$ ./replicate-data-local.sh -d path/to/dir -s
Downloading data for later import
You may want to download the data while in the office and restore it overnight to minimise disruption (or to provide to someone who doesn’t have integration access). First, do the download on your host, as the unzipping is a lot quicker when not run over NFS:
mac$ ./replicate-data-local.sh -u $USERNAME -n
Then follow the instructions above for importing using the
If you’re running out of disk space
If you get a curl error when restoring Elasticsearch data
Check the service is running:
dev$ sudo service elasticsearch-development.development start
Can’t take a write lock while out of disk space (in MongoDB)
You may see such an error message which will prevent you from creating or even dropping collections. So you won’t be able to replicate the latest data.
You will need to delete large Mongo collections to free up space before they can be re-imported. Follow this guide on how to delete them, and ensure that Mongo honours their removal.
Find your biggest Mongo collections by running:
dev$ sudo ncdu /var/lib/mongodb
You can re-run the replication but skip non-Mongo imports like MySQL if it’s already succesfully imported. Use
to see the options.
For example, to run an import but skip MySQL and Elasticsearch:
dev$ replicate-data-local.sh -q -e -d backups/2017-06-08 -s
Upgrading to Elasticsearch 2.4.6
If you have been running Elasticsearch 1.x.x in the development machine at some point the puppet run will install version 2.4.6. After this you will need to remove some plugins and the re-import the data in order for it to work with this version. In the
govuk-puppet/development-vm you can run the script
fix-elasticsearch-2.4.sh. This will remove any incompatible plugins and also delete the data from the previous version. After this you can replicate the Elasticsearch data again by running
./replicate-data-local.sh -u $USERNAME -m -p -q -t from the
More about Development VM
- Can't connect to Elasticsearch in VM
- Can't connect to Mongo in VM
- Content store times out in VM
- Fix issues with installing Ruby gems using Bundler
- Fix issues with vagrant-dns
- Fix low disk space in development
- Fix NFS errors in VM
- Increase RAM on the VM
- Problems fetching packages in VM
- Run an application in the VM
- Send a test email via Notify
- SSH into GOV.UK servers from the VM
- SSH into your VM directly
- Unable to mount VirtualBox shared folders
- Using `bowl` fails with bundler error