Problem
Recurring issue during HA failover: slave node corruption when rsync syncs from primary.
Config files become corrupted with null bytes, breaking services. Examples:
/etc/config/firewall → fw4 fails, VRRP reports "Operation not permitted"
/etc/nginx/mime.types → nginx won't start, shows binary data
- Files in rsync temp dir:
/usr/share/keepalived/rsync/etc/nginx/mime.types
Symptoms
- Slave behaves incorrectly: activates interfaces incorrectly
- VRRP can't communicate with master (packets visible in tcpdump, but kernel blocks)
- Secondary filesystem reports improper unmount: "FAT-fs: Volume was not properly unmounted"
Root Cause
Overlayfs/tmpfs corruption during concurrent rsync writes. File system secondary didn't unmount cleanly.
Proposed Fix
- Reduce rsync update frequency from current to 5-10 minute intervals
- Move temporary sync files from disk to tmpfs
- Investigate overlayfs mount state during failover
Problem
Recurring issue during HA failover: slave node corruption when rsync syncs from primary.
Config files become corrupted with null bytes, breaking services. Examples:
/etc/config/firewall→ fw4 fails, VRRP reports "Operation not permitted"/etc/nginx/mime.types→ nginx won't start, shows binary data/usr/share/keepalived/rsync/etc/nginx/mime.typesSymptoms
Root Cause
Overlayfs/tmpfs corruption during concurrent rsync writes. File system secondary didn't unmount cleanly.
Proposed Fix