Controller.loadJsonObject(path, true) returned {} for a missing file, so
loadPlugins' catch never ran and a plugin whose web build was expected but
missing was not reported (the second half of #1046; #1047 fixed the path).
With throwOnMissing set, every error now throws; without it only a missing
file is ignored, so a malformed JSON file throws again as before #1047.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BGUULoiHNbAegwMK4PSjHJ
* Add a request to mark hosts and the controller as needing a restart
Not every plugin arrives through PluginInstallRequest. Plugins checked
out into external_plugins or installed by hand on the machine never
flag a restart, so the web UI shows nothing and tooling that acts on
the flag never fires.
HostMarkRestartRequiredRequest and ControllerMarkRestartRequiredRequest
set config.restartRequired, guarded by the existing restart
permissions. They are exposed as "Mark Restart Required" in the control
dropdown of the host and controller pages, hidden once the flag is set,
and as mark-restart-required ctl commands. The mark shows up on the
next system metrics poll.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Cover the mark-restart-required ctl commands
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Not every plugin arrives through PluginInstallRequest. Plugins checked
out into external_plugins or installed by hand on the machine never
flag a restart, so the web UI shows nothing and tooling that acts on
the flag never fires.
HostMarkRestartRequiredRequest and ControllerMarkRestartRequiredRequest
set config.restartRequired, guarded by the existing restart
permissions. They are exposed as "Mark Restart Required" in the control
dropdown of the host and controller pages, hidden once the flag is set,
and as mark-restart-required ctl commands. The mark shows up on the
next system metrics poll.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
On Windows this test throws "EPERM: operation not permitted, rename
<path>" presumably because Windows prevents concurrent renames. This
does not affect the test outcome, so ignore the error(s).
These are used outside of the integration tests, but including the
integration index file causes the very slow setup of the contorller
and host. Avoid the integration setup costs when running commands
such as `pnpm mocha test/lib` by moving these utility functions to
the common test functions.
Without a transport Winston will print annoying "Attempt to write logs
with no transports" warnings all over the place. Setup a transport in
tests that cause logging to suppress this.
The behaviour of creating the config file if it's missing while
convenient causes confusion and problems when the tools are run
in the wrong folder. They end up creating a shadow installation
and misdirect the user with nonsensical error messages, instead
pointing to the real problem which is that the tool can't find
the configuration needed to actually run.
Require a config file to be present at startup and refuse to run if this
file is missing. To support the odd use-case of creating the config
when missing a --create-config flag is added. To conveniently generate
a config file to edit out from the "config create" command is added.
Ctl ended up calling its main class for "Control" for historical
reasons, along with its config file and plugin entrypoint.
The plugin entrypoint was renamed to ctl in 808eda0513. This
finishes that renaming by also renaming the main class, default config
file, config-control command, and main entrypoint.
The config-control command is renamed to config to be consistent with
the host and controller's config commands.
The term "control" is still kept in the config and message locations as
a common specifier for both ctl and web_ui.
_waitForRconReadyOrExit resolved with an Error to signal the process
exited before RCON came up, which is an odd shape for a promise. It now
resolves when RCON is ready and rejects otherwise, with the error emitted
with the exit, or "RCON connection lost" when there was none.
Observable behaviour is unchanged. sendRcon rejects with the same
messages as before and stop ignores the rejection, as stopping is done
once the process is gone.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The instance name on a log line is whatever the host sent, so a host can
attribute a line for one of its own instances to the wrong name. Compare
it with the name the controller has on record and replace it when they
differ.
Host connections can live for days, so the warning suppression expires
after five minutes instead of latching for the life of the connection,
and the bogus instance and misaligned name warnings share it so the two
together cannot spam the log.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Waiting for exit alongside rcon-ready made sendRcon reject with the flat
"RCON connection lost" for every case, including the unexpected exits
that used to reject with the EnvironmentError saying why the server went
away, as events.once rejects on error events.
Have the paired wait capture the error emitted with the exit and hand it
back, so sendRcon throws that when there is one and falls through to the
old message when there is not, which is the kill path where no error is
emitted. stop() ignores the error as its job is done once the process is
gone.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The fieldChanged handler repeated the hook invoke in each branch, and
factorio.shutdown_timeout had no invoke at all, so plugins were never
told about a change to it. Do the field specific side effect, then
invoke the hook once for every field.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
handleDebugDumpWsRequest sets clusterio_ignore_dump on the current
socket so the WsServer send wrapper does not echo the dumper's own
frames back into debugEvents. The flag was only set on the connector
"connect" event, but WsServerConnector.continue() installs a fresh
socket and emits "resume", not "connect", so a resumed session lost
the flag.
After a resume the next outgoing frame was echoed to debugEvents,
ws_dumper sent a DebugWsMessageEvent, the wrapper echoed that too,
and so on: infinite synchronous recursion ending in a RangeError:
Invalid string length from JSON.stringify in link.ts sendEvent. That
uncaught error made handleUnhandledErrors exit the controller.
Repro: admin sends DebugDumpWsRequest, the control connection drops
and resumes (e.g. a network blip), controller crashes on resume.
Set the flag on "resume" as well as "connect".
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Cooldude2606 <25043174+Cooldude2606@users.noreply.github.com>
Log messages share the root cause with the save details updates fixed in
the previous commit: the controller trusted host supplied instance
fields. LogMessageEvent only declares level and message in its schema,
but the Ajv instance is built without removeAdditional and the schema
does not set additionalProperties, so instance_id and instance_name sent
by a host survive validation and pass through fromJSON untouched. A host
could tag a log line with an instance assigned to another host and have
the controller store and serve it as that instance's output.
Lift the assignment check out of handleInstanceSaveDetailsUpdatesEvent
into a private method and use it from both handlers.
Strip the instance fields instead of dropping the line. A host racing a
reassignment does emit instance log lines for an instance the controller
has already moved, and dropping those loses real output exactly when a
migration is going wrong. The line is still genuine output from the host,
so it is recorded as a host level message, which grants the host nothing
it could not already log directly. Warn once per connection so a
misbehaving host cannot flood the log.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
_watchServerLogActions dereferenced two regex results with ! and
threw on null. The name regex /^([^ ]+)/ returns null when a line
classified as an action starts with a space, and the BAN reason
regex /\. Reason: (.+)\.$/ returns null for a [BAN] line with no
". Reason: X." tail. The output listener runs synchronously off the
stdout LineSplitter, so the throw becomes an uncaughtException and
takes the whole host down with every instance on it.
Repro: instance with enable_save_patching off and enable_script_commands
on; a control client sends /sc print("2000-01-01 00:00:00 [BAN] evil").
The raw print reaches stdout, parses as a dated BAN action, and the
reason deref throws.
Skip the line when the name regex misses and default the ban reason
to empty when the tail is absent. Apply the same name guard to
_watchPlayerPromote. The autosave ![1] in server.ts is safe because
its filter regex already guarantees the capture.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
A permission function runs during ingress permission checking, before
the request body is validated. UserBulkImportRequest.permission read the
raw, unvalidated message.data and threw a bare Error for an unknown
importType. That error is not a PermissionError, so _processMessage
rethrew it to the connector "message" listener, which emitted "error" on
the connector. Controller-side connectors have no "error" listener, so
Node rethrew and handleUnhandledErrors exited the process.
Repro: a control client with client-side validateSent disabled sends
UserBulkImportRequest with importType "foo"; the client gets a normal
RequestError response and the controller then logs a fatal uncaught
exception and exits.
Fix both layers: _processMessage now logs and drops an unexpected error
from the permission check (control connections only, where a response
was already sent) instead of rethrowing, and UserBulkImportRequest
throws a PermissionError for an unknown importType so it is denied
cleanly. A genuine PermissionError still denies silently as before.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
sendRcon waits for the rcon-ready event when RCON has not connected yet,
which is the same root cause as the stop() hang. events.once rejects on
error events, but _watchExit skips emitting error while the state is
stopping, and kill() sets that state before sending SIGKILL. exportData
does exactly that, so an RCON call racing the export never settles and
the request stays pending until the session closes.
Wait for exit as well, which leaves the existing RCON client check to
reject the call with "RCON connection lost" instead. Both call sites now
share a helper as the paired wait is identical.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Drop the special case that deferred terminating the connection issuing
the delete, so deleting yourself behaves the same as revoke-token and
every session of the deleted user is closed before the handler returns.
Set tokenValidAfter to the current time in createUser instead of 0.
Deleting a user and recreating it under the same name left tokens issued
before the deletion valid, as the check is iat < tokenValidAfter. Doing
it in createUser covers every path that makes a user. A token signed in
the same second as creation is still accepted, so create-admin and the
other production paths sign after the user exists and are unaffected.
The integration harness signed the control token at import time, before
bootstrap create-admin created the user. Sign it in the root before hook
instead and reuse it in link.js.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
ControlConnection keeps the User object it authenticated with and checks
permissions against it. Deleting the user removed the record but left
its sessions open, so a deleted user kept all its permissions until the
WebSocket closed.
Repro: create "victim" with role 0 and connect as it. Another admin
sends UserDeleteRequest("victim"). Victim can still send
UserCreateRequest("mallory") and UserUpdateRolesRequest("mallory", [0]).
Terminate the deleted user's connections like revoke-token does. When
a user deletes itself, defer terminating the requesting connection so
the response is sent first.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The config changed handler ran updateFactorioSettings and
updateFactorioWhitelist as soon as the field changed. While the instance
is starting, stopping, creating_save or exporting_data the server is not
ready for RCON, so the update threw and was only logged. The three list
update events pushed from the controller had the same problem, except
Link._processEvent swallowed the rejection and the update was lost
without a trace.
Add _waitForSettledStatus, which resolves once notifyStatus lands on
running or stopped. notifyStatus is the only place _status changes and
notifyExit always ends on stopped, so a waiter cannot hang. The settings,
whitelist and list updates now go through it.
Nothing is applied when the instance settles on stopped, because there is
no server to apply it to. factorio.settings is written to
server-settings.json and the lists to the list files on the next start,
and enableWhitelist is read from the config when the next Instance is
constructed.
The plugin hook now runs immediately for the two config fields instead of
being chained onto the update. Every other field already invokes it right
away, and waiting for the instance to settle would hold the notification
back for the whole startup.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The close handler removed the log subscription transport from the global
lib logger, but updateLogSubscriptions adds it to the controller's
clusterLogger. Every control connection that subscribed to logs stayed
attached to clusterLogger after closing, keeping the connection alive and
running every later cluster log line through its filter.
handleDebugDumpWsRequest added a new debugEvents listener on each call
and overwrote ws_dumper, so the close handler only removed the last one.
Repeated requests stacked listeners, triggered MaxListenersExceededWarning
after 11 calls, and left the earlier listeners behind. Ignore the request
if the connection is already dumping.
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
handleInstanceSaveDetailsUpdatesEvent trusted the instanceId in the
event and in each SaveDetails without checking the instance was assigned
to the sending host. InstanceSaveDetailsUpdatesEvent has host in its src
list and validateIngress only checks the source host id, not the
payload, so any authenticated host could corrupt another instance's
controller-side save records.
Repro: a host owning nothing sends the event with a SaveDetails for
instance 1002 (on another host) to inject a bogus save, or sends empty
updates with instanceId 1002 to make the delete-missing sweep wipe that
instance's save list.
Guard both paths: bail if event.instanceId names an instance not
assigned to this host, and drop the whole event if any update is for an
instance not assigned here. Matches handleInstanceStatusChangedEvent's
bogus-update warning.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
stop() waits for the rcon-ready event when RCON has not connected yet,
but nothing emits it once the process is gone, so the promise never
settles and the stop request stays pending until the session closes.
Wait for exit as well and drop the other listener when either fires.
Repro: send InstanceExportDataRequest to a stopped instance, then
InstanceStopRequest while the status is exporting_data. Instance.stop()
sees the server as running and calls server.stop(), then exportData
kills the process before RCON is up. The export finishes and the
instance reports stopped, but the stop request never gets a response.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
assignInstance sets instance.assigned_host to the new host before
sending InstanceAssignInternalRequest. If the host fails that request
the controller kept the new host id while the host never added the
instance. Assigning to the same host again then returned early as a
no-op. Repro: make the host's instances directory read-only, assign
instance 3001 to host 10 (EACCES on mkdir), make it writable, assign
again (ok, nothing happens), start (instance does not exist on host).
Now a failed send leaves the instance unassigned and rethrows.
InstanceCreateRequest applied instance.assigned_host from the request
config, creating an instance assigned to a host that never heard of
it. Repro: InstanceCreateRequest({"instance.id": 4001, "instance.name":
"c1", "instance.assigned_host": 10}). Reject a non-null value with the
same error the config set requests use.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The fieldChanged handler in the Instance constructor ran
updateFactorioSettings and updateFactorioWhitelist with only a finally
attached. A rejection from either went unhandled, and the host exits on
unhandled rejections, taking every instance on it down.
Two ways to hit it against a real host:
- Start an instance and immediately toggle factorio.enable_whitelist.
updateFactorioWhitelist calls sendRcon while the server state is still
"new" and it throws "Expected state running,stopping but state is new".
- Set factorio.settings tags to [1] on a running instance. applyAsConfig
called tag.replace outside its try block and threw
"tag.replace is not a function".
Catch and log both on the instance logger before invoking the hook, and
convert tags with String() so non-string entries no longer throw.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Two writes to the same file at the same time shared one temporary name, so the second write clobbered the first's temporary and the first rename left the second call with nothing to rename, failing with ENOENT. The host does exactly that when it saves an instance config while the instance saves it too.
The temporary now carries a random tag and is created exclusively, so concurrent writers never touch each other's file. Since the temporary is always new, the mode is applied on creation and the chmod for reused leftovers goes away. A failed rename removes its temporary instead of leaving it for the next write to overwrite.
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
The metrics endpoint fans out a gather to every connected host, so
anyone who could reach the controller could read cluster statistics and
put load on all hosts at will. Scraping now needs a user token holding
the new core.controller.metrics permission. The token is accepted via
Authorization: Bearer as well as X-Access-Token because Prometheus can
only send the former. Concurrent scrapes share one gather so hosts see
at most one metrics request at a time.
The three connection gauges in controller.ts now no-op before the
controller exists, like the other gauges there, so the endpoint can be
tested against the mock controller.
Closes#1009
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Resolves the ESM conflicts in patch.ts, Instance.ts and host.ts. The
module gathering in Instance now goes through patch.loadModules with
the instance's loadedPlugins, and imports use explicit extensions.
Review changes: the existing-output check only treats ENOENT as
"does not exist" and rethrows anything else, a --zip option keeps the
output as a zip file for transfer (extracting stays the default), the
--scenario option describes that it takes a path and what the default
is, and the docs note that IPC is not available outside Clusterio.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Rethrow errors other than ENOENT and ENOTDIR when reading the package
spec as a directory, mark the tests that run npm as slow so FAST_TEST
skips them, and use ./external_plugins/my_plugin in the README examples.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Plugins with only host, instance or ctl code have no reason to ship a
webpack build, yet the plugin list showed "Missing dist/web/manifest.json"
as an error for them and the controller logged a warning on every start.
Decide whether a build is expected from the plugin declaration, using the
same rule the create tool uses to decide when to generate one: a web or
controller entrypoint, or any config fields. Plugins outside that rule get
no bundle path and no error, the web UI skips loading them, and the plugin
pages show them as having no web module instead of an error or a prompt to
reload.
Fixes#1004
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
The mod pack's Factorio version ends up in mod-settings.dat. Factorio
refuses to read that file when the version is newer than the game and
falls back to default settings, which can wreck the save. Check the
versions in syncMods before touching the mods folder.
Closes#1006
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
