Chef Infra Server Prerequisites
The following is a detailed discussion of the prerequisites for every installation of the Chef Infra Server. See Install Chef Infra Server for installation instructions.
The following table lists the commercially-supported platforms for Chef Infra Server:
|Amazon Linux 2|
|Oracle Enterprise Linux|
|Red Hat Enterprise Linux|
|SUSE Linux Enterprise Server|
The following platforms are not tested by Chef Software:
- Any Linux or UNIX distribution that is not listed as a Foundational platform.
- Microsoft Windows
- 32-bit architectures
Read the guidance around capacity planning for information about how to choose the right topology for the Chef Infra Server.
All machines in a Chef Infra Server deployment have the following hardware requirements. Disk space for standalone and backend servers should scale up with the number of nodes that the servers are managing. A good rule to follow is to allocate 2 MB for each node. The disk values listed below should be a good default value that you will want to modify later if/when your node count grows. Fast, redundant storage (SSD/RAID-based solution either on-prem or in a cloud environment) is preferred.
- 64-bit CPU architecture
- CPU support for SSE4.2 extensions (Xeons starting in 2007 and Opterons in 2012)
- 4 total cores (physical or virtual)
- 8 GB of RAM or more
- 5 GB of free disk space in
- 10 GB of free disk space in
For a high availability deployment:
- Three backend servers; as many frontend servers as required
- 1 x GigE NIC interface (if on premises)
- 4 cores (physical or virtual)
- 4GB RAM
- 20 GB of free disk space (SSD if on premises, Premium Storage in Microsoft Azure, EBS-Optimized GP2 in AWS)
- 2 cores (physical or virtual)
- 8GB RAM
- 50 GB/backend server (SSD if on premises, Premium Storage in Microsoft Azure, EBS-Optimized GP2 in AWS)
Before installing the Chef Infra Server, ensure that each machine has the following installed and configured properly:
- Hostnames — Ensure that all systems have properly configured hostnames. The hostname for the Chef Infra Server must be a FQDN, have fewer than 64 characters including the domain suffix, be lowercase, and resolvable. See Hostnames, FQDNs for more information
- FQDNs — Ensure that all systems have a resolvable FQDN
- NTP — Ensure that every server is connected to NTP; the Chef Infra Server is sensitive to clock drift
- Mail Relay — The Chef Infra Server uses email to send notifications for various events; a local mail transfer agent should be installed and available to the Chef server
- cron — Periodic maintenance tasks are performed using cron
- git — git must be installed so that various internal services can confirm revisions
- Apache Qpid — This daemon must be disabled on CentOS and Red Hat systems
- Required users — If the environment in which the Chef Infra Server will run has restrictions on the creation of local user and group accounts, ensure that the correct users and groups exist before reconfiguring
- Firewalls and ports — If host-based firewalls (iptables, ufw, etc.) are being used, ensure that ports 80 and 443 are open.
- Browser — Firefox, Google Chrome, Safari, or Internet Explorer (versions 9 or better)
- Chef Infra Client communication with the Chef Infra Server The Chef Infra Server must be able to communicate with every node that will be configured by Chef Infra Client and every workstation that will upload data to the Chef Infra.
UIDs and GIDs
The installation process for the Chef Infra Server requires the use of
at least two user and group identifiers (UIDs and GIDs). These are used
to create the
opscode-pgsql users and their default
The creation of required user and group identifiers is done automatically during the installation process for the Chef Infra Server; however, the following user and group accounts may be created in advance of installing the Chef Infra Server if specific UIDs and GIDs are preferred. The user and group must be created as a pair to satisfy reconfiguration requirements.
- A local user account named
opscodeunder which services will run
- A local user account named
opscode-pgsqlthat is used by PostgreSQL
- A group account for each user account, one named
opscodeand the other named
opscode-pgsqlunder which services will run
If the UID and GID of
opscode-pgsql do not match on both
backend Chef Infra Server machines, a high availability configuration
will not run correctly.
The embedded Chef Infra Server cookbooks can handle two cases:
opscode-pgsqluser and group not found on the new server
opscode-pgsqluser and group found on the new server
Having only the group and not the corresponding users present during a chef-server-ctl reconfigure is unsupported and may lead to an error in the reconfiguration run.
To determine the current range of IDs, run the following command:
grep -E '(UID|GID)' /etc/login.defs
The defaults for CentOS and Red Hat Enterprise Linux systems look like this:
UID_MIN 500 UID_MAX 60000 GID_MIN 500 GID_MAX 60000
If the defaults have been changed for any reason, and if that change
would result in less than 2 UID/GIDs being available to the
/etc/login.defs with changes to make at least 2 more
UIDs and GIDs available for association. The currently used ID ranges
for UIDs and GIDs can be found in
opscode-pgsql user and group identifiers exist
prior to installing the Chef Infra Server, the Chef Infra Server
installation process will use the existing identifiers instead of
To allow access to your Chef Infra Server on ports 80 and 443 via the iptables firewall, issue the following command with root privileges:
iptables -A INPUT -p tcp -m multiport --destination-ports 80,443 -j ACCEPT
Note that you will need to make use of a tool such as iptables-persistent to restore your iptables rules upon reboot.
On RHEL and CentOS versions 7 and above, the FirewallD firewall is enabled by default. Issue the following command with root privileges to open ports 80 and 443:
firewall-cmd --permanent --zone public --add-service http && firewall-cmd --permanent --zone public --add-service https && firewall-cmd --reload
While UFW is installed on Ubuntu, it is not enabled by default. However, if you wish to use a UFW-based firewall on your Chef Infra Server, issue the following command with root privileges to open ports 80 and 443:
ufw allow proto tcp from any to any port 80,443
On CentOS and Red Hat Enterprise Linux systems, SELinux is enabled in
enforcing mode by default. The Chef Infra Server does not have a profile
available to run under SELinux. In order for the Chef Infra Server to
run, SELinux must be disabled or set to
To determine if SELinux is installed, run the following command:
If a response other than
"Permissive" is returned,
SELinux must be disabled.
To set SELinux to
Permissive mode, run:
and then check the status:
On Ubuntu systems, AppArmor is enabled in enforcing mode by default.
Chef products do not have a profile available to run under AppArmor. In
order for the Chef products to run, AppArmor must set to
mode or disabled.
To determine if AppArmor is installed, run the following command:
To install AppArmor, run the following command:
sudo apt-get install apparmor-utils -yes
If a response other than
"0 processes are in enforce mode" or
"0 profiles are in enforce mode." is returned, AppArmor must be set to
Complaining mode or disabled.
To set AppArmor to
Complaining mode, run:
sudo aa-complain /etc/apparmor.d/*
Or to disable AppArmor entirely, run:
sudo invoke-rc.d apparmor kill sudo update-rc.d -f apparmor remove
and then check the status:
On CentOS and Red Hat Enterprise Linux systems, the Apache Qpid daemon is installed by default.
To determine if Apache Qpid is installed, run the following command:
rpm -qa | grep qpid
If Apache Qpid is installed, a response similar to the following is displayed:
To disable Apache Qpid run:
service qpidd stop
chkconfig --del qpidd
Periodic maintenance tasks are performed on the Chef Infra Server
servers via cron and the
/etc/cron.d directory. With certain CentOS 6
configurations, an additional step is required to install crontab:
yum install crontabs
Enterprise Linux Updates
The Chef Infra Server requires an x86_64 compatible systems
architecture. When the Chef Infra Server is installed on Red Hat
Enterprise Linux or CentOS, run
yum update prior to installing the
Chef Infra Server. This will ensure those platforms are fully compatible
with this requirement.
Unless you intend to operate the Chef Infra Server in IPv6 mode, you
should disable ipv6 in the system’s
/etc/hosts file by commenting out
or removing all references to IPv6 addresses like “::1” or
Without these changes, a Chef Infra Server install intended to run in ipv4 mode will mistakenly only start the postgres service on the ipv6 loopback address of “::1” rather than the ipv4 loopback address of 127.0.0.1. This will make further progress through an initial reconfiguration impossible.
The hostname for the Chef Infra Server may be specified using a FQDN or
an IP address. This hostname must be resolvable, be 64 characters or
less, and be lowercase. For example, a Chef Infra Server running in a
production environment with a resolvable FQDN hostname can be added the
DNS system. But when deploying Chef Infra Server into a testing
environment, adding the hostname to the
/etc/hosts file is enough to
ensure that hostname is resolvable.
FQDN Hostnames When the hostname for the Chef Infra Server is a FQDN be sure to include the domain suffix. For example, something like
mychefserver.example.com(and not something like
IP Address Hostnames When the Chef Infra Server is run in IPv6 mode, a hostname specified using an IP address must also be bracketed (
[ ]) or the Chef Infra Server will not be able to recognize it as an IPv6 address. For example:
bookshelf['vip'] = "hostname.example.com" # If we're getting weird. Change to IPv6 bookshelf['vip'] = "[2001:db8:85a3:8d3:1319:8a2e:370]"
api_fqdn setting can be added to the private-chef.rb file (it is
not there by default). When added, its value should be equal to the FQDN
or IP address for the service URI used by the Chef Infra Server. Then
configure the same value for the
bookshelf['vip'] setting prior to
installing the Chef Infra Server. FQDNs must always be in lowercase.
api_fqdn "chef.example.com" or
Use the following sections to verify the hostnames that is used by the Chef Infra Server.
To verify if a hostname is a FQDN
To verify if a hostname is a FQDN, run the following command:
If the hostname is a FQDN, it will return something like:
If the hostname is not a FQDN, it must be configured so that it is one.
To verify the FQDN is all lowercase
To verify if the alphabetic parts of a FQDN are all lowercase, run the following command:
hostname -f | grep -E '^([[:digit:]]|[[:lower:]]|\.|-|_)+$' && echo yes
If the hostname is all lowercase, it will return something like:
If the hostname’s alphabetic parts are not all lowercase, it must be configured so that they are.
To verify a hostname is resolvable
To verify is a hostname is resolvable, run the following command:
If the hostname is resolvable, it will return something like:
To change a hostname
In some cases, the hostname for the Chef Infra Server needs to be updated. The process for updating a hostname varies, depending on the platform on which the Chef Infra Server will run. Refer to the manual for the platform or contact a local systems administrator for specific guidance for a specific platform. The following example shows how a hostname can be changed when running Red Hat or CentOS:
sudo hostname 'mychefserver.example.com'
echo "mychefserver.example.com" | sudo tee /etc/hostname
To add a hostname to /etc/hosts
If a hostname is not resolvable, refer to a local systems administrator
for specific guidance on how to add the hostname to the DNS system. If
the Chef Infra Server is being into a testing environment, just add the
/etc/hosts. The following example shows how a hostname can
be added to
/etc/hosts when running Red Hat or CentOS:
echo -e "127.0.0.2 `hostname` `hostname -s`" | sudo tee -a /etc/hosts
CNin a certificate to be no longer than 64 characters.
The Chef Infra Server server uses email to send notifications for various events:
- Password resets
- User invitations
- Failover notifications
- Failed job notifications
Configure a local mail transfer agent on the Chef Infra Server using the steps appropriate for the platform on which the Chef Infra Server is running.
The Chef Infra Server requires that the systems on which it is running be connected to Network Time Protocol (NTP), as the Chef Infra Server is particularly sensitive to clock drift. For Red Hat and CentOS 6:
yum install ntp
chkconfig ntpd on
service ntpd start
apt-get install ntp
Chef Infra Client
The Chef Infra Server server requires that every node that is under
management by Chef also have an accurate clock that is synchronized very
closely with the clock on the Chef Infra Server. If the clocks are not
synchronized closely, the authentication process may fail when the
clocks are out-of-sync by more than 15 minutes. A failure will trigger a
401 Unauthorized response similar to:
[Tue, 01 Nov 2011 16:55:23 -0700] INFO: *** Chef 11.X.X *** [Tue, 01 Nov 2011 16:55:23 -0700] INFO: Client key /etc/chef/client.pem is not present - registering [Tue, 01 Nov 2011 16:55:24 -0700] INFO: HTTP Request Returned 401 Unauthorized: Failed to authenticate as ORGANIZATION-validator. Synchronize the clock on your host. [Tue, 01 Nov 2011 16:55:24 -0700] FATAL: Stacktrace dumped to /var/chef/cache/chef-stacktrace.out [Tue, 01 Nov 2011 16:55:24 -0700] FATAL: Net::HTTPClientException: 401 "Unauthorized"
In this situation, re-synchronize the system clocks with the Network Time Protocol (NTP) server, and then re-run Chef Infra Client.
By default, accounts required by the Chef Infra Server are created during setup. If your environment has restrictions on the creation of local user and group accounts that will prevent these accounts from being created automatically during setup, you will need to create these accounts.
The following group accounts are required:
|The group name under which services will run.|
The following user accounts are required:
|The user name under which services will run.|
|The user name for PostgreSQL. (This is only required on the back end servers in a high availability setup.)|