Regenerate certificates in split installations

Expand menu Collapse menu

Delete and recreate the certificate authority in a split installation
Regenerate compile master certificates
Regenerate agent certificates
Regenerate infrastructure certificates in a split installation

Regenerating certificates and security credentials—both private and public keys—created by the built-in PE certificate authority can help ensure the security of your installation in certain cases.

Important: A split installation includes any configuration where the PE infrastructure components are split across more than one node, such as the traditional split installation or a standalone PE-PostgreSQL installation. A "database node" in these instructions is your PuppetDB and/or standalone PE-PostgreSQL node. If the console runs on your master node, skip instructions for the console node, except as indicated.

The process for regenerating certificates varies depending on your goal.


If your goal is to...	Do this...
Upgrade to the intermediate certificate architecture introduced in Puppet 6.0.	Complete these tasks in order: Delete and recreate the certificate authority in a split installation Regenerate compile master certificates, if applicable Regenerate agent certificates
Fix a compromised or damaged certificate authority.
Fix a compromised compile master certificate or troubleshoot SSL errors on compile masters.	Regenerate compile master certificates
Fix a compromised agent certificate or troubleshoot SSL errors on agent nodes.	Regenerate agent certificates
Specify a new DNS alt name or other trusted data.	Regenerate infrastructure certificates in a split installation

Delete and recreate the certificate authority in a split installation

Recreate the certificate authority only if you're upgrading to the new certificate architecture introduced in Puppet 6.0, or if your certificate authority was compromised or damaged beyond repair.

Warning: Replacing your certificate authority invalidates all existing certificates in your environment. Complete this task only if and when you're prepared to regenerate certificates for both your infrastructure nodes and your entire agent fleet.

Perform these steps logged in as root.

Back up these directories:
Master
- /etc/puppetlabs/puppet/ssl/
- /etc/puppetlabs/orchestration-services/ssl
Note: In a standalone PE-PostgreSQL configuration, back up /etc/puppetlabs/puppetdb/ssl/ and /opt/puppetlabs/server/data/console-services/certs/ on your master.
Database node
- /etc/puppetlabs/puppet/ssl/
- /etc/puppetlabs/puppetdb/ssl/
- /opt/puppetlabs/server/data/postgresql/9.6/data/certs/
Console node
- /etc/puppetlabs/puppet/ssl/
- /opt/puppetlabs/server/data/console-services/certs/

Stop these PE services:

Master

puppet resource service puppet ensure=stopped
puppet resource service pxp-agent ensure=stopped
puppet resource service pe-puppetserver ensure=stopped
puppet resource service pe-bolt-server ensure=stopped
puppet resource service pe-orchestration-services ensure=stopped

Note: In a standalone PE-PostgreSQL configuration, stop pe-console-services and pe-puppetdb on your master.

Database node

puppet resource service pe-puppetdb ensure=stopped
puppet resource service pe-postgresql ensure=stopped

Console node

puppet resource service pe-console-services ensure=stopped

Remove certificate files and the cached catalog, and then rebuild certificates on each infrastructure node, in the order specified:
Master
```
rm -rf /etc/puppetlabs/puppet/ssl/*
rm -f /opt/puppetlabs/puppet/cache/client_data/catalog/<CERTNAME>.json
```
```
puppet infrastructure configure --no-recover
```
Note: In a standalone PE-PostgreSQL configuration, PuppetDB on the master fails when you run puppet infrastructure configure because the database isn't reachable on the standalone node. Accept the failure, configure the database node, and then re-run puppet infrastructure configure --no-recover on your master.
Database node
```
rm -rf /etc/puppetlabs/puppet/ssl/*
rm -f /opt/puppetlabs/puppet/cache/client_data/catalog/<CERTNAME>.json
```
```
puppet infrastructure configure --no-recover
```
Console node, if applicable
```
rm -rf /etc/puppetlabs/puppet/ssl/*
rm -f /opt/puppetlabs/puppet/cache/client_data/catalog/<CERTNAME>.json
```
```
puppet infrastructure configure --no-recover
```
Run Puppet on your master, database node, and console, as applicable: puppet agent -t

Regenerate compile master certificates

Regnerate compile master certificates to fix a compromised certificate or troubleshoot SSL errors on compile masters, or if you recreated your certificate authority.

Before you begin

Configure Puppet Server with allow-subject-alt-names in the certificate-authority section of ca.conf.

If you did not recreate your certificate authority, you must log into your master of masters as root and remove the compile master certificate: puppetserver ca clean --certname <COMPILE_MASTER_HOSTNAME>
Log into the compile master as root and back up the /etc/puppetlabs/puppet/ssl/ directory:
```
cp -r /etc/puppetlabs/puppet/ssl/ /etc/puppetlabs/puppet/ssl_bak/
```

Stop the Puppet agent, Puppet Server, and PXP agent services.

puppet resource service puppet ensure=stopped
puppet resource service pe-puppetserver ensure=stopped
puppet resource service pxp-agent ensure=stopped

Delete the compile master SSL directory: rm -rf /etc/puppetlabs/puppet/ssl
Remove the compile master cached catalog: rm -f /opt/puppetlabs/puppet/cache/client_data/catalog/<CERTNAME>.json
Adjust DNS alt names and trusted data, as needed.
Re-start the Puppet agent service: puppet resource service puppet ensure=running
After the agent starts, it automatically generates keys and request a new certificate from the Puppet CA.
Log into your master of masters as root and sign the compile master's certificate request: puppetserver ca sign --certname <COMPILE_MASTER_HOSTNAME>
Log into your compile master as root and run Puppet: puppet agent -t

PE performs a full catalog run, and the compile master resumes its role in your deployment.

Regenerate agent certificates

Regenerate agent certificates to fix a compromised certificate or troubleshoot SSL errors on agents, or if you recreated your certificate authority.

Unless otherwise indicated, perform these steps on the agent node that you're regenerating certificates for.

If you did not recreate your certificate authority, you must log into your master and clear the cert for the agent node: puppetserver ca clean --certname <CERTNAME>
Back up the /etc/puppetlabs/puppet/ssl/ directory.
If something goes wrong, you might need to restore these directories so your deployment remains functional.

Stop the Puppet agent and PXP agent services.

puppet resource service puppet ensure=stopped
puppet resource service pxp-agent ensure=stopped

Delete the agent SSL directory.
- On *nix nodes: rm -rf /etc/puppetlabs/puppet/ssl
- On Windows nodes, use the Administrator confdir to delete the $confdir\ssl directory.
Remove the agent's cached catalog.
- On *nix nodes: rm -f /opt/puppetlabs/puppet/cache/client_data/catalog/<CERTNAME>.json
- On Windows nodes, use the Administrator confdir to delete the $client_datadir\catalog\<CERTNAME>.json file.
Re-start the Puppet agent service: puppet resource service puppet ensure=running
After the agent starts, it automatically generates keys and request a new certificate from the Puppet CA.
If you aren't using autosigning, sign each agent node's certificate request using the console's request manager, or from your master:
```
puppetserver ca list 
puppetserver ca sign --certname <NAME>
```
From the console or command line, run Puppet on the node.

The agent performs a full catalog run, restarts the PXP agent service, and resumes its role in your PE deployment.

Regenerate infrastructure certificates in a split installation

Regenerate infrastructure certificates to specify a new DNS alt name or other trusted data.