CockroachDB With GSSAPI Deployed via Systemd

This tutorial bridges the gap between the documentation to deploy CockroachDB via systemd and actual steps to deploy CockroachDB with GSSAPI and systemd.

Artem Ervits

CORE ·

Feb. 18, 22 · Tutorial

Likes (3)

Comment

Save

8.4K Views

Articles Covering Topics on CockroachDB and Kerberos

I find the topic of Kerberos very interesting and my colleagues commonly refer to me for help with this complex topic. I am by no means an expert at Kerberos, I am however familiar enough with it to be dangerous. That said, I've written multiple articles on the topic which you may find below:

CockroachDB With MIT Kerberos
CockroachDB With Active Directory
CockroachDB With MIT Kerberos and Docker Compose
Executing CockroachDB table import via GSSAPI
CockroachDB With SQLAlchemy and MIT Kerberos
CockroachDB With MIT Kerberos Cert User Authentication
CockroachDB with Django and MIT Kerberos
CockroachDB With Kerberos and Custom Service Principal Name (SPN)
Simplifying CockroachDB Kerberos Architecture With a Load Balancer
CockroachDB With MIT Kerberos Using a Native Client
CockroachDB With Mixed Kerberos and Certificates Authentication

Motivation

Systemd has become a standard approach for deploying Linux services. We have documentation to deploy CockroachDB via systemd, however, we do not have steps documented to deploy CockroachDB with GSSAPI and systemd. This tutorial attempts to bridge that gap.

Generally, all of the steps should be similar to my previous articles, specifically the very first article in the series. I also had to refer to my article on using the native CockroachDB binary to access a cluster with GSSAPI as we've made that possible since I'd written the first article. And yes, I blog to refer to my articles like documentation, by I digress.

The only thing that will change is starting the service via a systemd config. GSSAPI in CockroachDB relies on an environment variable KRB5_KTNAME that is not available to a process started by systemd. In systemd, we have to add the environment variable to the systemd config. We have a docs issue outstanding to document this, thank you Fabio for doing your research!

High-Level Steps

Create a KDC and populate with principals
Install CockroachDB
Generate certs for a secure installation
Start CockroachDB via systemd
Confirm KRB5_KTNAME environment variable is read in the process
Access CockroachDB as a KDC user to verify GSSAPI works

Step by Step Instructions

Populate KDC With a List of Principals

    
   sudo kadmin.local
kadmin.local:  
  addprinc postgres/node.example.com@EXAMPLE.COM
  addprinc postgres/node@EXAMPLE.COM
  addprinc postgres/127.0.0.1@EXAMPLE.COM
  addprinc postgres/localhost@EXAMPLE.COM
  addprinc carl@EXAMPLE.COM

Still, in the kadmin interface, dump the principals to a keytab:

    
     ktadd -k keytab postgres/127.0.0.1@EXAMPLE.COM
  ktadd -k keytab postgres/node.example.com@EXAMPLE.COM
  ktadd -k keytab postgres/node@EXAMPLE.COM
  ktadd -k keytab postgres/localhost@EXAMPLE.COM

Verify the Keytab is Populated Correctly

    
  

   [root@node ~]# ktutil
ktutil:  read_kt keytab
ktutil:  list
slot KVNO Principal
---- ---- ---------------------------------------------------------------------
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2           postgres/127.0.0.1@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2    postgres/node.example.com@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2                postgres/node@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM
  2           postgres/localhost@EXAMPLE.COM

  

Install CockroachDB

    
   COCKROACH_VERSION=v21.1.5
wget -qO- https://binaries.cockroachdb.com/cockroach-$COCKROACH_VERSION.linux-amd64.tgz | tar  xvz
sudo cp -i cockroach-$COCKROACH_VERSION.linux-amd64/cockroach /usr/local/bin/
cockroach version

Create CockroachDB Certs

    
  

   HOSTNAME="localhost node.example.com node 127.0.0.1"
mkdir certs my-safe-directory
cockroach cert create-ca --certs-dir=certs --ca-key=my-safe-directory/ca.key
cockroach cert create-node $HOSTNAME --certs-dir=certs --ca-key=my-safe-directory/ca.key
openssl x509 -in certs/node.crt -text | grep "Subject Alternative Name" -A 1
cockroach cert create-client root --certs-dir=certs --ca-key=my-safe-directory/ca.key
  

Start Node(s) Using Systemd

I am working on this tutorial with a single VM and I'm going to defer to using a single node CockroachDB instance with systemd instead of spinning up three processes. Other than using start-single-node and skipping the initialization step, everything should be the same.

    
  

   mkdir /var/lib/cockroach
useradd cockroach
mkdir /var/lib/cockroach/certs
mv certs/node.* /var/lib/cockroach/certs/
cp certs/ca.crt /var/lib/cockroach/certs/
mv keytab /var/lib/cockroach/
chown -R cockroach.cockroach /var/lib/cockroach
  

Create Systemd Config

    
  

   [Unit]
Description=Cockroach Database cluster node
Requires=network.target
[Service]
Type=notify
WorkingDirectory=/var/lib/cockroach
Environment="KRB5_KTNAME=/var/lib/cockroach/keytab"
ExecStart=/usr/local/bin/cockroach start-single-node --certs-dir=certs --advertise-addr=node.example.com:26257 --cache=.25 --max-sql-memory=.25
TimeoutStopSec=60
Restart=always
RestartSec=10
StandardOutput=syslog
StandardError=syslog
SyslogIdentifier=cockroach
User=cockroach
[Install]
WantedBy=default.target

  

Save as /etc/systemd/system/securecockroachdb.service.

Notice the Environment property, that's how we override the environment variable from our docs.

Environment="KRB5_KTNAME=/var/lib/cockroach/keytab"

Start CockroachDB via Systemd

systemctl start securecockroachdb

Confirm the KRB5_KTNAME environment variable is set by first getting the PID:

ps aux | grep cockroach

    
     PID TTY      STAT   TIME COMMAND
 5573 ?        Rsl    0:07 /usr/local/bin/cockroach start-single-node --certs-dir=certs --advertise-addr=node.example.com:26257 --cache=.25 --max-sql-memory=.25 LANG=en_US.UTF-8 PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin NOTIFY_SOCKET=/run/systemd/notify HOME=/home/cockroach LOGNAME=cockroach USER=cockroach SHELL=/bin/bash KRB5_KTNAME=/var/lib/cockroach/keytab
[root@node ~]# ps aux | grep cockroach
cockroa+  5573  7.4  1.4 926204 114908 ?       Ssl  12:35   0:08 /usr/local/bin/cockroac start-single-node --certs-dir=certs --advertise-addr=node.example.com:26257 --cache=.25 --max-sql-memory=.25
root      5598  0.0  0.0  12528   972 pts/0    R+   12:37   0:00 grep --color=auto cockroach

View the In-Process Environment Variable

strings /proc/5573/environ

    
  

   LANG=en_US.UTF-8
PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin
NOTIFY_SOCKET=/run/systemd/notify
HOME=/home/cockroach
LOGNAME=cockroach
USER=cockroach
SHELL=/bin/bash
KRB5_KTNAME=/var/lib/cockroach/keytab

  

There are a few more ways to get the same info:

    
   cat /proc/5573/environ | tr '\0' '\n'
ps e -ww -p 5573

Enable Enterprise License for GSSAPI

cockroach sql --certs-dir=certs --host=node

    
   SET CLUSTER SETTING cluster.organization = 'Acme Company';
SET CLUSTER SETTING enterprise.license = 'xxxxxxxxxxxx';

Enable GSSAPI

    
   SET cluster setting server.host_based_authentication.configuration = 'host all all all gss include_realm=0';

Add a SQL user named Carl:

    
   CREATE USER carl;
GRANT ALL ON DATABASE defaultdb TO carl;

Verify GSSAPI Works

kinit as Carl:

    
   kinit carl
klist

    
   Ticket cache: KEYRING:persistent:0:0
Default principal: carl@EXAMPLE.COM

Valid starting       Expires              Service principal
07/09/2021 12:45:31  07/10/2021 12:45:31  krbtgt/EXAMPLE.COM@EXAMPLE.COM

Connect to CockroachDB Using Carl as Username

    
   /usr/local/bin/cockroach sql \
 --certs-dir=certs --host=node --url "postgresql://carl@node:26257/defaultdb?sslmode=verify-full&sslrootcert=certs/ca.crt"

    
  

   [root@node ~]# /usr/local/bin/cockroach sql \
>  --certs-dir=certs --host=node --url "postgresql://carl@node:26257/defaultdb?sslmode=verify-full&sslrootcert=certs/ca.crt"
#
# Welcome to the CockroachDB SQL shell.
# All statements must be terminated by a semicolon.
# To exit, type: \q.
#
# Server version: CockroachDB CCL v21.1.5 (x86_64-unknown-linux-gnu, built 2021/07/02 03:57:09, go1.15.11) (same version as client)
#
# Enter \? for a brief introduction.
#
carl@node:26257/defaultdb>
  

Verify GSSAPI Is Still Enabled

Connecting to CockroachDB again using root user cert:

cockroach sql --certs-dir=certs --host=node

    
   root@node:26257/defaultdb> SHOW CLUSTER SETTING server.host_based_authentication.configuration;
  server.host_based_authentication.configuration
--------------------------------------------------
  host all all all gss include_realm=0
(1 row)

That's it for today!

CockroachDB Kerberos (protocol)

Published at DZone with permission of Artem Ervits. See the original article here.

Opinions expressed by DZone contributors are their own.

Related

Trending