Features
The smaller tools under each dashboard surface - not a second pass through Overview and Live.
Health grade + restart advice
Letter grade, plain reasons when it is not Strong, and Safe / Caution / Wait restart advice. Long uptime plus worse GC can suggest a quiet maintenance window. WatchTower does not restart the server for you.
Fix inbox ranking
Watching and Scanning feed a ranked inbox with one plain next step per issue. No giant scheduled audit dump.
- MSPT spike after restart3m ago#1
- Disk runway under 14 days1h ago#2
- Client-only jar on serverYesterday#3
World pressure
Entity, item, and chunk census that spots item storms, mob spikes, and unattended loaders.
- Dropped items1,842
- Chicken412
- Sheep286
- Other318
- Items Storm risk82%
- Mobs Elevated54%
- Chunks Loaders on71%
Join / pack sync clinic
Failed joins map to named mod diffs on Session. Player-safe copy of the fix - read-only, no jar downloads.
Live vitals charts
TPS, MSPT, players, heap, CPU, and host charts while you watch - including honest hosted-panel metrics.
- facts.jsonredacted
- brief.mdplain English
- evidence/logs + crash
Support pack redaction
Build a redacted zip (facts, brief, evidence) for a helper or mod author. Discord copy presets stay consistent with the pack.
Spark lag proof
Optional Spark companion turns a profile into what ate the tick. Deep Spark workspace is Alpha.
ContinuousOBBCollider.collideMany8.2%GC / JVM + RAM advice
GC pause share of wall time, JVM flags profile, and RAM advice that uses your host or container memory limit - not a one-size guess.
Heap is the pressure — GC pauses are quiet.
Crash fingerprints
Crash reports grouped and explained in plain English, with context from nearby logs.
a3f2…c919b10…e2aMod inventory + Modrinth hints
Installed jars, conflicts, Modrinth lookup hints, and mod log errors with Active / Reviewed. Modrinth never downloads jars for you.
- CCreate6.0.4 · createServer
- JJust Enough Items19.21.0 · jeiServer
- OOptiFineconflict · optifineClient-only
Pack / jar drift
Checksum baseline drift and high-confidence client-only jars surfaced on Issues.
e7c1…90afb44d…12ce- ~
create-6.0.0.jarchecksum drift - +
sodium-extra-0.6.jaradded since baseline - !
optifine-1.21.jarclient-only on server
Soft jar disable / enable
Rename a mod jar to `*.jar.disabled` so it skips the next boot (or rename it back). Filter All / Enabled / Disabled. High world risk asks you to confirm first. Admins only - no delete.
create-6.0.0.jarcreate-6.0.0.jar.disabledMods → Configs
Edit files under the server `config/` folder from the dashboard. TOML gets a form when WatchTower can parse it; otherwise you edit the raw text. Saves create a backup and support undo. Admins only.
- bool
enableHopper=true - num
tickRate=20 - num
maxDepth=8
External kill / OOM
Host OOM killer vs panel force-kill when there is no crash report - plus the right fix path.
Killed process java · total-vm 8GAbrupt stop · SIGKILL windowSilent script fails
KubeJS, CraftTweaker, datapack, and /reload errors that never crash still become Issues.
- KubeJSrecipes.js:142 · Failed to parse recipe3m
- CraftTweaker/reload aborted · script error12m
- Datapackfunction tag missing · #minecraft:tick1h
- KubeJSserver_scripts · event never fired2h
Log tail
latest.log triage in the dashboard so you are not bouncing to the host panel for every line.
- 14:02:11[Server]Done (42.1s)! For help, type "help"
- 14:17:58[Watch]TPS 19.8 · heap 61%
- 14:18:03[Watch]MSPT 48.2 warn
- 14:18:04[Issue]ranked #1 · MSPT spike after restart
- 14:18:41[Join]djinn disconnected · channel mismatch
- 14:19:11[Join]clinic matched · named mod diffs
Schedule + load trends
Busy vs quiet hours so you plan restarts around real load. Times follow the timezone you set in the dashboard.
Activity / incident stories
Pulls lag spikes, crashes, and missed backups into one readable incident thread.
- Lag14:18MSPT spiked to 48.2After restart · players still joining
- Crash14:31Server process exitedSame window as the lag spike
- Backup15:02Scheduled backup missedNo folder change since 12:00
Storage space map
A treemap of what is using disk space. Drill into World, Logs, Mods, or Backups.
Storage + disk runway
Dimension storage scan plus roughly how many days of disk left - not just a percent full.
Weekly ops digest
Local rollup of grade, crashes, disk, and MSPT trend with one next action. Stays on your host.
Config audit
Read-only keep / tweak / why for server.properties and startup flags.
- Keep
view-distance=10Fine for this pack - Tweak
max-tick-time=-1Hides stuck ticks - Why
G1GC flags=partialMissing Aikar set
Backup health
See whether local backups look present and fresh, then verify zip/tar.gz integrity. Optional test restore only under `watchtower/restore-verify/` - never into the live world.
world/Freshpanel/Stalecloud/UnknownSpark Map
Pan and zoom chunk heat from the selected Spark profile. Click a chunk for details.
Startup watch
First-minutes and boot health when the process comes up.
- JVM8.2s
- Mods22.4s
- World11.1s
- Ready…
Sources freshness
Poller freshness and what data pull is next so you know if Watching is current.
- ops-cacheWatching12s
- live samplesCurrent4s
- latest.logTailed18s
- SparkIdle—
Named accounts + audit log
Owner / admin / viewer logins, optional Minecraft player link on the side rail, Sign out, and a Settings audit log of account and settings changes.
- settings.savedjinn · 2m ago
- account.invitedjinn · 1h ago
- role.changemaya · Yesterday
Theme + accent
Light, Dark, Black, or System, plus an accent color. Saved per signed-in account.
Secure login + optional 2FA
Login required by default; optional 2FA for the dashboard.
Help Center
In-app wiki with the same guides as the public GitHub wiki.
- StartDashboard Overview
- TriageIssues
- SessionJoin Clinic
Disaster-recovery CLI + viewer
Matching CLI jar and browser viewer when Minecraft will not stay up.
$ java -jar watchtower-cli.jar
watchtower-cli 0.4 · DR mode
> compose support --redact
wrote support-pack.zip
facts · brief · evidence (redacted)Open the demo on sample data, or get the jar on Modrinth.