Disk Full Emergency: No SSH, No CWP Access, Server Partially Down – Full Rescue

Disk Full Emergency – No SSH, No CWP Access, Server Partially Down (Rescue Guide)This post documents a real production incident where the root filesystem ( / ) reached 100% usage, causing:

• No access to CWP7 panel
• SSH connections failing or hanging
• WordPress admin showing “There has been a critical error”
• MariaDB / Postfix / services failing
• System unable to write logs or temporary files

Below are the exact steps used to recover the server safely and permanently.Symptoms Observed

• CWP panel unreachable
• SSH timeout or login failure
• journalctl errors: “No space left on device”
• MariaDB errors writing temporary tables
• Postfix queue errors
• WordPress admin critical error

Step 1: Confirm Disk Is Full
 If usage is 100%, the system is in a critical state and must be freed immediately.Step 2: Free Emergency Space (Safe First Actions)Clear temporary directories (safe to delete):
 Re-check disk usage:
 This often restores enough space to regain temporary control.Step 3: Identify What Is Using Disk SpaceLocate large directories:
 In this case, the issue was found here:
 Cache size had grown to tens of gigabytes.Step 4: Remove Runaway WordPress CacheImportant:
Do NOT delete uploads or databases.
Remove cache only.
 Verify disk space is restored:
 At this point, the server should become responsive again.Step 5: Restart Critical Services
 Confirm:

• SSH works
• CWP panel loads
• WordPress admin loads

Root CauseThe root cause was disk-based WordPress caching (W3 Total Cache):

• Disk page cache enabled
• No size limits
• No purge failure protection
• Traffic / bots caused cache explosion

Disk-based caching silently filled the root filesystem.Permanent Fix (Highly Recommended)

1. Disable Disk Cache Completely

• Disable all disk-based caching in WordPress
• Never use disk page cache on production servers

1. Use RAM-Based Caching

• Page Cache → Memcached
• Database Cache → Redis
• Object Cache → Redis

1. Enforce Memory Limits

Redis:
 Memcached:
 Critical Safety Net (Do Not Skip)Add a cron job to guarantee cleanup even if plugins fail:
 This single line prevents future disk-full disasters.Optional: Browser CacheEnable Browser Cache in WordPress and set long cache headers for CSS, JS, images.
This improves performance without using disk or RAM.Final Result

• Server fully recovered
• CWP panel accessible
• SSH stable
• WordPress admin restored
• Disk usage normalized
• Cache architecture hardened

Key Takeaways

• Disk-based caching can silently destroy servers
• Always prefer RAM-based caching
• Always enforce memory limits
• Always keep an emergency cleanup cron

This rescue procedure was tested in a real production outage.