CentOS 7下PostgreSQL 9.6/9.5初始化及服务启动故障求助

Let's break down how to resolve the two key issues you're facing with PostgreSQL 9.5/9.6 on CentOS 7 for your Netbox setup:

Why postgresql96-setup is Missing #

That initialization script is part of PostgreSQL's official RPM packages (hosted in their dedicated repository). If you installed PostgreSQL from CentOS's default base repo, this script won't exist—those packages use a different structure and don't include this helper tool.

Option 1: Install from PostgreSQL's Official Repo (to get the setup script) #

If you want to use the postgresql96-setup tool, reinstall PostgreSQL from the official repo:

# Uninstall existing PostgreSQL packages first
yum remove postgresql96-server postgresql96
# Add the official PostgreSQL 9.6 repo for CentOS 7
yum install https://download.postgresql.org/pub/repos/yum/9.6/redhat/rhel-7-x86_64/pgdg-centos96-9.6-3.noarch.rpm
# Install server and client packages
yum install postgresql96-server postgresql96
# Now you can run the setup script as intended
/usr/pgsql-9.6/bin/postgresql96-setup initdb

Option 2: Manually Initialize the Database (Keep Your Current Install) #

Since you already have initdb in your /usr/pgsql-9.6/bin/ directory, you can initialize the database directly without the helper script:

1. Verify the postgres system user exists (it should be created automatically during installation):

   id postgres
   # If missing, create it:
   useradd postgres
   

2. Switch to the postgres user (never run PostgreSQL as root—it's a security risk and causes permission issues):

   su - postgres
   

3. Initialize the database cluster, specifying the default data directory:

   /usr/pgsql-9.6/bin/initdb -D /var/lib/pgsql/9.6/data
   

4. Fix directory permissions (critical for startup):

   # Switch back to root
   exit
   chown -R postgres:postgres /var/lib/pgsql/9.6/data
   chmod 700 /var/lib/pgsql/9.6/data
   

Fixing the Failed PostgreSQL Service Startup #

When systemctl shows the service failed, logs are your best friend—they'll tell you exactly what's wrong.

1. Check PostgreSQL's Local Logs #

The default log directory for PostgreSQL 9.6 on CentOS 7 is /var/lib/pgsql/9.6/data/pg_log/. Look at the most recent log file for errors:

tail -n 50 /var/lib/pgsql/9.6/data/pg_log/*.log

Common issues you might see:

• Permission errors: The postgres user can't access the data directory (we fixed this above, but double-check)
• Port conflict: Another service is using port 5432 (PostgreSQL's default). Verify with:

  netstat -tulpn | grep 5432
  

• Configuration mistakes: Typos in postgresql.conf (e.g., invalid listen_addresses) or incorrect rules in pg_hba.conf

2. Use Systemd's Journal for Detailed Startup Logs #

Run this command to get a granular breakdown of why the service failed:

journalctl -xeu postgresql-9.6.service

3. SELinux Troubleshooting #

CentOS 7 has SELinux enabled by default, which can block PostgreSQL even if filesystem permissions are correct:

• Test if SELinux is the issue by temporarily disabling it (don't leave it disabled long-term):

  setenforce 0
  

• If the service starts after disabling SELinux, apply the correct SELinux context to the data directory:

  semanage fcontext -a -t postgresql_db_t "/var/lib/pgsql/9.6/data(/.*)?"
  restorecon -Rv /var/lib/pgsql/9.6/data
  # Re-enable SELinux
  setenforce 1
  

4. Start the Service Properly #

Use systemctl (the recommended tool for CentOS 7) to manage the service:

systemctl start postgresql-9.6
# Enable it to start automatically on boot
systemctl enable postgresql-9.6

内容的提问来源于stack exchange，提问作者uk_newt