PostgreSQL 9.4迁移数据目录后启动超时失败求助

Hey there, let's work through this startup timeout issue you're facing after moving your PostgreSQL data directory. This is a common problem with a few key culprits to check, especially on a production server. Here's what to look into:

1. Verify Directory Permissions (Most Likely Culprit) #

PostgreSQL runs as the postgres user, so it needs full read/write/execute access to the new data directory and all its parent directories. Even if you used rsync, permissions might not have been preserved correctly or parent directories might block access:

• Check the ownership of the new data directory:

  ls -ld /home/databasepostgre/pgsql/9.4/data
  

  The output should show postgres postgres as the owner and group.
• If not, fix the ownership recursively:

  sudo chown -R postgres:postgres /home/databasepostgre/pgsql/9.4/data
  

• Also, ensure parent directories have at least execute permission for the postgres user (so it can traverse into the data directory):

  sudo chmod o+x /home/databasepostgre/pgsql/9.4/
  sudo chmod o+x /home/databasepostgre/pgsql/
  sudo chmod o+x /home/databasepostgre/
  

2. Check SELinux Contexts #

Most production servers have SELinux enabled, and moving the data directory breaks the security context that allows PostgreSQL to access the files.

• First, test if SELinux is the issue by temporarily disabling it (don't leave this off permanently):

  sudo setenforce 0
  

  Then try starting the service again:

  sudo systemctl start postgresql-9.4.service
  

  If it starts successfully, fix the SELinux context permanently:

  sudo chcon -R --reference=/var/lib/pgsql/9.4/data /home/databasepostgre/pgsql/9.4/data
  sudo restorecon -R /home/databasepostgre/pgsql/9.4/data
  

  Re-enable SELinux:

  sudo setenforce 1
  

  Then start the service again.

3. Check Systemd Service Overrides #

Some PostgreSQL systemd service files hardcode the PGDATA path, which overrides the setting in postgresql.conf.

• Inspect the service file to see if there's an explicit PGDATA environment variable:

  cat /usr/lib/systemd/system/postgresql-9.4.service
  

  Or check for any override files:

  ls /etc/systemd/system/postgresql-9.4.service.d/
  cat /etc/systemd/system/postgresql-9.4.service.d/*.conf
  

• If you find a line like Environment=PGDATA=/var/lib/pgsql/9.4/data, update it to your new path, then reload systemd:

  sudo systemctl daemon-reload
  

  Then try starting the service.

4. Look at PostgreSQL's Own Logs #

The systemd timeout logs are vague—PostgreSQL's internal logs will give you specific errors (like permission denied, missing files, or stale lock files). Check the log directory inside your new data folder:

ls /home/databasepostgre/pgsql/9.4/data/pg_log/
cat /home/databasepostgre/pgsql/9.4/data/pg_log/<latest-log-file>.log

Look for lines like FATAL: could not open directory "pg_xlog": Permission denied or FATAL: lock file "postmaster.pid" already exists—these will point directly to the problem.

5. Check for Stale Lock Files #

If PostgreSQL was shut down improperly, a postmaster.pid file might be left in the new data directory, preventing a fresh start:

• Check if the file exists:

  sudo ls /home/databasepostgre/pgsql/9.4/data/postmaster.pid
  

• If it does, stop any lingering PostgreSQL processes (if any) and delete the file:

  sudo pkill -u postgres
  sudo rm /home/databasepostgre/pgsql/9.4/data/postmaster.pid
  

  Then try starting the service again.

Start with these checks—permissions and SELinux are the most common causes here. Let me know if you find a specific error in the PostgreSQL logs that needs more clarification!

内容的提问来源于stack exchange，提问作者ian ajah