Improvements
4- SSH reconnect failures are now tracked in a 5-minute sliding window (max 10) instead of permanently locking out the session after a transient burst.
- Backend startup now re-attaches in-flight migrations via RecoverActiveMigrations, preventing orphaned migration sessions after a restart.
- Pre-migration disk space preflight check prevents starting a migration when the destination lacks sufficient free space.
- Outer panic guard in pipeline.Execute prevents a single failing step from crashing the entire migration goroutine.
Bug Fixes
7- CheckpointManager is now initialised on every Run() call so resumable migrations actually resume from the last checkpoint.
- Empty-user PHP-FPM health check no longer spams logs when no domains are present on the source server.
- Plesk email discovery now queries psa.mail JOIN domains directly (Method 0), reducing noise from fallback heuristics.
- rndc reload exit 1 now falls back to reconfig+retry before marking DNS as failed.
- wp-config update fast-path skips the rewrite step when database name, username, and password hash are all preserved.
- MariaDB to MySQL 8 collation sanitizer now rewrites utf8mb4_uca1400_* collations to compatible MySQL 8 equivalents, preventing data load failures.
- DNS service now attempts rndc reconfig first and falls back to a full reload, eliminating single-shot DNS transition failures.