When /home fills up on a cPanel server, the system does not crash — it does worse. Mail bounces silently, MySQL refuses writes that involve temp files, AutoSSL skips renewals, cron jobs fail without alerting, and PHP returns half-written responses. The recovery is not just “delete some files” — the order matters because some operations require free space to succeed, and deleting the wrong thing first will dig the hole deeper.
What’s happening
cPanel puts user data, mail spool, and per-user MySQL temp directories on /home. When the partition hits 100%, three things break simultaneously: (1) mail cannot deliver to /home/<user>/mail, queues build up in /var/spool/exim; (2) MySQL cannot create on-disk temp tables on the affected partition; (3) AutoSSL writes its work files to /home/.cpanel/autossl/ and fails. Some of these recover automatically when space is restored; some need manual intervention.
How to verify
Confirm fullness and find the largest consumers:
# is it actually 100%?
df -h /home
df -i /home
# top 20 largest directories
du -sh /home/* 2>/dev/null | sort -hr | head -20
# top 20 largest files anywhere on /home
find /home -xdev -type f -size +100M -exec ls -lh {} \; 2>/dev/null | sort -k5 -hr | head -20
# specific common culprits
du -sh /home/*/logs /home/*/tmp /home/*/mail 2>/dev/null | sort -hr | head
# Exim queue blocked by no mail-spool write
exim -bpc
ls /var/spool/exim/input | wc -l
The fix
- Free space first, gently — find a non-critical large directory (a user’s
tmp/, an old backup in/home/backup/, a forgotten log archive). Delete it. The goal is to get above the 95% threshold so other services start working again. - Restart cpsrvd and Exim —
systemctl restart cpsrvd exim. Once disk has space, the daemons recover but they need a kick. Checkexim -bpcis decreasing. - Identify the real consumer — was this a one-time spike (a user uploaded a backup) or a slow growth (a log directory never rotates)?
du -sh /home/*over 24 hours of history (if you log it) shows the pattern. - Lower quotas if a user is the cause —
whmapi1 modifyacct user=<user> QUOTA=10000. Communicate before lowering; quotas below current usage make the user immediately over-quota. - Move data off
/home— if the server genuinely needs more space, the long-term fix is a larger/homeor a separate partition for backups./backupshould not be on/home.
Once recovered, set a monitor alert at 80% so the next time gives 24-hour warning. Stack Harbor handles capacity-restoration runbooks under cPanel/WHM management — start a conversation when disk fills are a recurring pattern.