Phase 8: 20+ new config keys + 14 admin commands (/playersync)
Config (JdbcConfig.java completely restructured into sections):
connection
host, port, use_ssl, user_name, password, db_name, table_prefix, Server_id
general
sync_world, sync_advancements, kick_when_already_online,
kick_message, kick_grace_period_ms, use_legacy_serialization,
item_placeholder_title_override, item_placeholder_description_override
save_triggers
auto_save_interval_minutes (0-1440, default 10)
save_on_dimension_change (default false)
save_on_death (default true)
save_on_respawn (default true)
sync_toggles
sync_inventory, sync_ender_chest, sync_xp, sync_effects,
sync_health_food, sync_curios, sync_accessories, sync_backpacks,
sync_cosmetic_armor, sync_refined_storage (all default true)
performance
heartbeat_interval_seconds (5-600, default 30)
peer_stale_threshold_seconds (10-3600, default 60)
join_poll_max_attempts (10-600, default 120)
join_poll_interval_ms (100-5000, default 500)
pool_stats_interval_minutes (0-1440, default 5)
hikari_pool_max_size (1-200, default 15)
hikari_leak_threshold_ms (2000-600000, default 25000)
safety
refuse_empty_inventory_write (default true) — enforced in writeSnapshotToDB
max_inventory_size_bytes (default 10 MB)
skip_saves_when_tps_below (0-20, default 0 = never)
observability
log_structured_json (future use)
log_rotation_size_mb (default 10)
log_rotation_max_files (default 5)
Wiring
- HeartbeatService reads heartbeat_interval_seconds at start.
- PoolStatsReporter reads pool_stats_interval_minutes (0 disables).
- doPlayerJoin poll uses join_poll_max_attempts + join_poll_interval_ms +
peer_stale_threshold_seconds.
- writeSnapshotToDB: refuse_empty guard + max_inventory_size_bytes guard
before core UPDATE. Both log via SyncLogger.dataLoss / .nbtAnomaly.
- Restore-side toggles: applyCuriosFromData, applyAccessoriesFromData,
applyCosmeticArmorFromData, doBackPackRestore, restoreRefinedStorageDisks
all short-circuit when their toggle is false.
Commands — new /playersync tree (perm level 2 required):
status — server id + heartbeat age + exec/Hikari stats + online
poolstats — log current stats immediately
flush [player] — force save all / one
info <player> — DB row metadata
dump <player> — dump full DB row to server log
resync <player> — clear synced tag + kick to force re-restore
wipe <player> confirm — DELETE all rows (DANGER, double-keyword required)
orphans — list stuck online=1 rows on dead peers
clearorphans [id] — clear orphans (global or by server_id)
peers — list peer servers with ALIVE/STALE/STOPPED tag
peerkill <id> — force-disable a zombie peer
cleanup — orphans + stale peers in one shot
reload — note about runtime reload scope
help — in-chat command reference
Every command logs to SyncLogger as ADMIN_<OP> for audit trail.
Infrastructure
- JDBCsetUp.executePreparedUpdateRet(String, Object...) returns rows-affected
for commands that need meaningful counts.
- VanillaSync.getExecutor() exposes the thread pool for read-only stats access
from admin commands (replaces reflection use in PoolStatsReporter eventually).
c7487196ec