Server Auto Restart

This page documents the current modules/alerts/Server Auto Restart/ module from the newer FioChat source.

This module is not only a restart-message broadcaster. The current source runs an internal scheduler, sends checkpoint warnings before the restart, executes final actions when the countdown reaches zero, writes restart state to disk, and can run follow-up actions after the server or plugin comes back online.

Configuration Files

• modules/alerts/Server Auto Restart/config.yml
• modules/alerts/Server Auto Restart/state.yml

config.yml is the main module configuration. state.yml is created automatically so the plugin can remember that a restart was triggered and that after actions may need to run on the next startup.

Legacy path still recognized by the core migration flow:

• modules/autorestart/config.yml

Module Toggle

The module master toggle is stored in modules/settings.yml:

modules:
  autorestart: true

The service reads modules.autorestart. If the key is missing, the service fallback is false; the current default module settings file explicitly writes it as true.

When the module is disabled, no scheduler is created, /autorestart reports that the module is disabled, and no checkpoint or restart action is executed.

Permission

Permission	Default	Detail
fiochat.autorestart.manage	op	Allows control of /fc autorestart <now|delay|stop|freeze> and /autorestart <now|delay|stop|freeze>. Console can also run the control command.

The command handler also accepts FioChat's internal admin permission as an alternative, but the module-specific node written in plugin.yml is fiochat.autorestart.manage.

Commands

Main command from plugin.yml:

/autorestart <now|delay|stop|freeze>

FioChat also routes:

/fc autorestart now
/fc autorestart delay <time>
/fc autorestart stop
/fc autorestart freeze

/fc autorestart now

Shows the current countdown state.

If the scheduler is stopped, the command reports that it is stopped. If it is frozen, the command reports the frozen remaining time. If a countdown is active, it shows the remaining duration and clock format.

In this source, now does not force an immediate restart. It inspects the active countdown.

/fc autorestart delay <time>

Replaces the next restart countdown with the provided duration.

Examples:

/fc autorestart delay 30s
/fc autorestart delay 5m
/fc autorestart delay 1h 20m

The service cancels active tasks, clears stopped/frozen state, and schedules a new cycle using that duration. The runtime minimum is 20 ticks, or 1 second.

/fc autorestart stop

Stops the active auto restart scheduler.

The service cancels all currently scheduled restart/checkpoint tasks, sets schedulerStopped: true, clears frozen state, and clears nextRestartAtMillis.

Stopping does not change the config. After a plugin reload or module reload, the config is read again and the scheduler can become active again.

/fc autorestart freeze

Freezes the countdown at its current remaining time.

The service reads remainingTicks(), cancels active tasks, sets schedulerFrozen: true, stores the remaining time in frozenRemainingTicks, and clears nextRestartAtMillis.

The current command handler provides freeze, but it does not provide a matching resume command. To continue, an admin usually uses delay <time> or reloads the module/plugin.

Duration Format

Config values and the delay command support:

10s
5m
1h 30m
6h
1w 2d

Recognized units:

Unit	Aliases	Value
tick	tick, ticks, t	1 tick
second	second, seconds, sec, secs, s	20 ticks
minute	minute, minutes, min, mins, m	1,200 ticks
hour	hour, hours, hr, hrs, h	72,000 ticks
day	day, days, d	1,728,000 ticks
week	week, weeks, w	12,096,000 ticks

If the input is only a number with no unit, the service treats the number as ticks, not seconds. For example, 600 means 600 ticks, or 30 seconds.

The parser accepts duration segments separated by whitespace. 1h 30m is valid. Input with non-whitespace characters between segments is rejected.

settings

Current default:

settings:
  debug-log: false
  initial-delay: "30s"
  interval: "6h"

settings.debug-log

Enables debug logging for the auto restart module.

The current default is false. When enabled, the service logs reload summaries and each successfully executed action.

The debug key is autorestart. It is useful for checking how many checkpoints were loaded, how many before/after actions were loaded, and which actions were actually executed at each stage.

settings.initial-delay

The delay after plugin load/reload before the first auto restart cycle starts.

The current default is 30s. The service converts it to ticks through the duration parser. The runtime minimum is 0 ticks.

The first restart happens after:

initial-delay + interval

With initial-delay: "30s" and interval: "6h", the first restart happens about 6 hours and 30 seconds after the service reloads. Broadcast checkpoints are calculated relative to that final restart time.

If this is set to 0s, the first cycle begins without an extra startup delay.

settings.interval

The duration of one auto restart cycle.

The current default is 6h. The service forces a minimum of 20 ticks. If the config value is invalid, the service fallback is 72,000 ticks, or 1 hour; however, the current default config writes 6h.

A checkpoint is only scheduled if its remaining duration is not greater than the active interval. For example, with interval: "10m", a 30m checkpoint is ignored because it cannot occur inside a 10-minute cycle.

After the final action runs, the service schedules the next cycle using settings.interval as long as the module is still enabled and the scheduler is neither stopped nor frozen.

before

Current default:

before:
  enabled: true
  actions:
    - "[BROADCAST] <red>Server is restarting now.</red>"
    - "[SOUND] minecraft:block.note_block.bell:1.0:0.8"
    - "[CONSOLE] save-on"
    - "[CONSOLE] save-all"
    - "[RESTART]"

This section runs when the countdown reaches zero.

before.enabled

Enables or disables the final pre-restart action section.

If this is false, the service treats before as empty and tries legacy fallback sections: before-restart, then restart, then final-actions. If all of them are empty, the countdown can still complete, but no final action from this section is executed.

before.actions

The ordered list of actions executed when restart time arrives.

Actions run from top to bottom. %time%, %ticks%, %seconds%, and %minutes% are still processed, but at the before-restart stage the remaining time passed to actions is 0.

Default action behavior:

• [BROADCAST] <red>Server is restarting now.</red> sends a chat announcement to all online players and console.
• [SOUND] minecraft:block.note_block.bell:1.0:0.8 plays the bell sound for all online players.
• [CONSOLE] save-on runs save-on as console.
• [CONSOLE] save-all runs save-all as console.
• [RESTART] runs FioChat's internal layered restart action.

Legacy Fallback Sections

The source still supports older section names:

• before-restart
• restart
• final-actions

Fallback is only used when before produces no actions. The order is before, before-restart, restart, then final-actions.

[RESTART]

[RESTART] has special behavior.

When executed, ModuleActionUtil tries:

1. Calling Bukkit.getServer().spigot().restart() through reflection.
2. Dispatching the console command restart.
3. Scheduling fallback Bukkit.shutdown().

If a restart attempt was made, fallback shutdown is scheduled 40 ticks later. If no restart attempt could be made, fallback shutdown is scheduled sooner.

This means [RESTART] is not just a command string. It is a layered restart mechanism intended to make the server actually leave the active process.

after

Current default:

after:
  enabled: false
  actions:
    - "[CONSOLE] say FioChat auto restart completed successfully."

This section runs after the server/plugin comes back online, not immediately before restart.

after.enabled

Enables or disables post-restart actions.

The current default is false. If disabled, the service does not keep an active after-action list. On reload, if a pending after-restart state exists but no after actions are active, the service clears the pending state.

after.actions

The ordered list of actions executed once after restart.

Before running before.actions, the service writes:

pending-after-restart: true
last-restart-at-millis: <timestamp>

On the next plugin load/reload, if pending-after-restart: true and after.actions are active, the service clears the pending state and schedules the after actions 40 ticks later.

Legacy Fallback Section

The source still supports after-restart if after produces no actions.

broadcast

Current default:

broadcast:
  messages:
    - ''
    - '<align:center><gold><bold>Server Auto Restart</bold></gold></align>'
    - '<align:center><dark_gray>(<smallfont:Alert Auto Restart>)</align>'
    - ''
    - '<align:center><gray>This server will be restart in: <white>%time%</white></align>'
    - '<align:center><gray>If the countdown reaches 0 you will be teleported to Hub.</gray></align>'
    - ''
  interval:
    - '30m'
    - '15m'
    - '5m'
    - '3m'
    - '1m'
    - '30s'
    - '15s'
    - '10s'
    - '5s'
    - '3s'
    - '2s'
    - '1s'

broadcast is the newer checkpoint-warning system. The service reads broadcast.interval, then creates a checkpoint action for each valid interval.

broadcast.messages

The chat-message template sent at every checkpoint interval.

Each line is automatically converted into a BROADCAST action with [prefix="false"] prepended. Empty lines are still sent as blank lines to players.

Placeholders such as %time% are replaced using the checkpoint's remaining time. At the 5m checkpoint, %time% becomes 5m; at the 30s checkpoint, it becomes 30s.

These template messages are merged with the interval-specific actions from broadcast.actions.<interval>.

broadcast.interval

The list of remaining times before restart when checkpoint actions should run.

The current default runs checkpoints at:

• 30m
• 15m
• 5m
• 3m
• 1m
• 30s
• 15s
• 10s
• 5s
• 3s
• 2s
• 1s

The service ignores intervals that are empty, invalid, zero/negative, or greater than settings.interval.

Checkpoints are scheduled with:

runAfter = initialDelay + (intervalCycle - checkpointRemaining)

For a 6-hour cycle and a 30-minute checkpoint, the checkpoint runs 5 hours and 30 minutes after the cycle begins, plus the initial delay for the first cycle.

broadcast.actions

Additional actions for a specific checkpoint.

The current config uses keys matching broadcast.interval, for example:

broadcast:
  actions:
    '30m':
      - '[SOUND] BLOCK_NOTE_BLOCK_PLING 1 2 master'
      - '[TITLE] <yellow><bold>!</bold> Auto Restart <bold>!</bold>\n<white>Server restart in <gold>%time%</gold>'

When the 30m checkpoint runs, the service executes:

1. All broadcast.messages as BROADCAST actions.
2. All actions in broadcast.actions.'30m'.

If no interval-specific actions exist, the checkpoint can still run as long as broadcast.messages is not empty.

checkpoints

Besides broadcast, the service still supports a legacy-style checkpoints section.

Supported shape:

checkpoints:
  5m:
    enabled: true
    actions:
      - "[BROADCAST] <yellow>Restart in %time%</yellow>"
      - "[SOUND] BLOCK_NOTE_BLOCK_PLING 1 2 master"

The service only reads checkpoints if broadcast produces no checkpoints at all. With the current default config, broadcast is used, not checkpoints.

Action Types

The action parser supports [TYPE] payload and several legacy prefixes such as BROADCAST: message.

[CONSOLE]

Runs a command as console.

Examples:

- "[CONSOLE] save-all"
- "[CONSOLE] say Restart soon"

A leading slash is removed automatically if present.

Token aliases: [COMMAND], [CMD], and legacy CONSOLE_COMMAND:.

[PLAYER]

Runs a command as every online player.

Example:

- "[PLAYER] spawn"

This action runs for all online players because auto restart actions are executed through executeForOnlinePlayers.

Legacy prefix: PLAYER_COMMAND:.

[BROADCAST]

Sends an announcement message to all online players and also to console when the message is not blank.

Example:

- "[BROADCAST] <red>Server restarts in %time%</red>"

Token alias: [CHAT]. Legacy prefix: BROADCAST:.

[MESSAGE]

Sends a chat message to all online players, but does not mirror the message to console like BROADCAST.

Token alias: [MSG]. Legacy prefix: MESSAGE:.

[ACTIONBAR]

Sends an actionbar to all online players.

Example:

- "[ACTIONBAR] <yellow>Restart in %time%</yellow>"

Legacy prefix: ACTIONBAR:.

[TITLE]

Sends a title to all online players.

Separator format:

[TITLE] title||subtitle||fadeIn||stay||fadeOut

Newline format is also supported. The default config uses \n, and the parser converts it into a real newline. If the payload contains a newline and no ||, the first line becomes the title and the second line becomes the subtitle. The default fade values for newline format are 10, 40, 10.

Examples:

- "[TITLE] <red>Restart Warning||<white>In %time%</white>||10||50||10"
- "[TITLE] <yellow>Auto Restart</yellow>\n<white>Server restart in %time%</white>"

[BOSSBAR]

Creates a temporary bossbar for all online players.

Format:

[BOSSBAR] title||color||style||progress||seconds

Behavior:

• color falls back to WHITE if invalid.
• style falls back to SOLID if invalid.
• progress is clamped from 0.0 to 1.0.
• seconds has a minimum of 1.

Example:

- "[BOSSBAR] <red>Restart in %time%</red>||RED||SOLID||1.0||5"

[SOUND]

Plays a sound for all online players.

Space-separated format:

[SOUND] SOUND_NAME volume pitch category

Colon-separated format:

[SOUND] minecraft:block.note_block.pling:1.0:1.0

The current default config uses both styles:

- "[SOUND] minecraft:block.note_block.bell:1.0:0.8"
- "[SOUND] BLOCK_NOTE_BLOCK_PLING 1 2 master"

If the sound cannot be resolved for the server version, the sound action is skipped.

[RESTART]

Runs the internal layered restart mechanism documented in the [RESTART] section.

The payload may be empty. This differs from most other action types, which usually require a payload.

Action Placeholders

Placeholders replaced in every auto restart action:

Placeholder	Value
%time%	Remaining time in readable format, such as 30m, 1m, or 5s.
%ticks%	Remaining time in ticks.
%seconds%	Remaining time in seconds, rounded down from ticks.
%minutes%	Remaining time in minutes, rounded down from ticks.

For checkpoints, the values use the checkpoint remaining time. For before and after, the service passes 0 remaining ticks.

PlaceholderAPI/FioChat Placeholders

FioChat provides these auto restart status placeholders:

Placeholder	Output
%fiochat_autorestart_seconds%	Remaining countdown in seconds. If no service is active, output is 0.
%fiochat_autorestart_formatted%	Remaining countdown in readable format, such as 1h 5m 3s.
%fiochat_autorestart_hhmmss%	Remaining countdown as compact HHMMSS.
%fiochat_autorestart_hh:mm_ss%	Remaining countdown as HH:MM SS.
%fiochat_autorestart_hh:mm ss%	Alias for HH:MM SS.
%fiochat_autorestart_lastrestart_seconds%	Time since the latest marked restart in seconds.
%fiochat_autorestart_lastrestart_formatted%	Time since the latest marked restart in readable format.
%fiochat_autorestart_lastrestart_hhmmss%	Time since the latest marked restart as compact HHMMSS.
%fiochat_autorestart_lastrestart_hh:mm_ss%	Time since the latest marked restart as HH:MM SS.
%fiochat_autorestart_lastrestart_hh:mm ss%	Alias for HH:MM SS.

lastrestart depends on lastRestartAtMillis, which is set when the before-restart stage runs or is restored from state.yml after startup.

state.yml

The automatic state file is:

modules/alerts/Server Auto Restart/state.yml

Important values:

pending-after-restart: true
last-restart-at-millis: 1710000000000

pending-after-restart

Marks that the before-restart stage reached its final point and that after-restart actions should run on the next load.

The service writes true before executing before.actions. On the next reload, if after actions are active, the service clears the flag to false and schedules after actions 40 ticks later.

last-restart-at-millis

The millisecond timestamp when the service marked the restart as pending.

This value is used by the lastrestart placeholders so the server can report how long it has been since the latest marked restart.

Runtime Flow

On module reload:

1. The service shuts down old tasks.
2. It reads modules.autorestart.
3. It reads settings.debug-log.
4. If the module is disabled or config is missing, reload stops.
5. It resets stopped/frozen state.
6. It reads settings.initial-delay.
7. It reads settings.interval.
8. It reads checkpoints from broadcast; if none are produced, it reads checkpoints.
9. It reads before; if empty, it tries legacy fallback sections.
10. It reads after; if empty, it tries after-restart.
11. It runs pending after-restart actions if state.yml requests them.
12. It schedules the first cycle.

When the countdown reaches zero:

1. The service writes state.yml as pending after restart.
2. It executes before.actions.
3. It sets lastRestartAtMillis.
4. It clears nextRestartAtMillis.
5. If the server does not actually stop and the scheduler is still valid, it schedules the next cycle.

Because the default before.actions contains [RESTART], the server normally restarts or shuts down before the next scheduled cycle matters.

Page note

This page is source-backed and includes a short callout so the content reads like a guide instead of plain reference text.