Property Reference

This document is regenerated from the physical configuration templates (*.properties) being deployed. The authoritative source of the templates is plantpulse-datalake-cli/config/templates, and the host's /etc/kopens/conf is bind-mounted to that location. Which template is rendered where is determined by config/render.map. The authoritative source for loading priority is plantpulse-core/src/plantpulse/core/engine/utils/PropertiesUtils.java.

If keys have been added, deleted, or renamed, the template—not this document—is the authoritative source. Do not fill in this document manually; extract it again from the template. The "Default" values in each table are exactly as written in the template, and ${PP_XXX} are placeholders substituted with environment variable values during rendering.

Value Flow: Source to Destination

Item	Location	Permissions	How to Change
Secrets — DB passwords · API keys	/etc/kopens/plantpulse-platform.env	0600	bin/passwd.sh (Change server account → Update sidecar → Re-render → Restart as one sequence)
Node-specific values — PP_MASTER_IP · Kafka advertised address · Image tag pins	/etc/kopens/platform.node.env	0600	Edit directly, then bin/restart-datalake.sh
Settings — Ports · Image tags · Locale · Data paths	.env in repo (Commit .env.example only)	—	Edit directly, then restart
«Lines» of Engine Settings — postgresql.conf parameters, Kafka retention, Aster TSDB heap	/etc/kopens/conf/<파일>.template	—	Edit template → pd config render → pd restart <서비스>

Directly editing component config files will cause changes to disappear on next startup

pd start renders the template every time and overwrites the configuration location of each component. Even if you open and edit cassandra.yaml, postgresql.conf, or kafka.properties, they will revert to their original state upon the next startup. The place to make changes is the template → How to change settings

Do not merge secrets and node values into a single file

Secret files are copied to app nodes during split installation, whereas node files must not be copied. Merging them will leak one box's PP_MASTER_IP and image tag pins to other boxes. Only one format is supported: VAR=값 — if export is appended, env_file: will fail to parse it.

Standard Template Change Procedure

Modify values on the host and reflect them into pd inside the container.

vi /etc/kopens/conf/<템플릿>                      # 1. 호스트의 템플릿을 고칩니다

docker exec plantpulse-datalake pd doctor         # 2. 검사 (선택)
docker exec plantpulse-datalake pd config diff    # 3. 무엇이 달라지는지 봅니다
docker exec plantpulse-datalake pd config render  # 4. 실제로 씁니다
docker exec plantpulse-datalake pd restart <서비스>  # 5. 재시작
docker exec plantpulse-datalake pd status         # 6. 확인

pd config render only rewrites the configuration file and does not restart the service. Until a restart occurs, the new configuration exists only on disk while the old process remains active. This is why steps 4 and 5 are a single set.

Configuration File Loading Priority

Web applications (server / batch / cep / sql / data-gateway) search for configuration files in the following order. The implementation is PropertiesUtils.readExternalFirst().

1. Directory specified as a -Dpp.conf.dir system property (constant PropertiesUtils.CONF_DIR_SYSTEM_PROPERTY = "pp.conf.dir")
2. ${catalina.base}/../config — The config/ directory of the module
3. Web app classpath (defaults included in WAR) — fallback

External files are read as UTF-8. If external configurations are loaded successfully, the following is printed in the startup log:

Properties loaded from external conf: <absolute-path>

If reading an external file fails, an error log is recorded, and it falls back to the classpath.

Logging configurations (log4j2.xml) are also externalized through the same chain (self-seeding — if no external file exists, WAR defaults are copied to the module config/log4j2.xml once and then used). When applied, Logging reconfigured from external conf: <absolute-path> is printed in the startup log. Changes to log levels are applied by modifying module config/log4j2.xml and restarting, which is preserved even after WAR redeployment. Refer to the configuration and deployment guide for details.

Caution: Do not directly modify internal web app files (server/webapps/ROOT/WEB-INF/classes/, etc.) — they will disappear upon WAR redeployment.

${ENV:기본값} Substitution Occurs Twice

The ${PP_XXX} / ${PP_XXX:기본값} notation in templates is interpreted at two different points in time.

Timing	Agent	Action
At Startup	pd config render	Substituted with sidecar (plantpulse-platform.env · platform.node.env) and container environment. Order: Environment variable → :기본값 → Preserve original. Nested placeholders are not supported
At Runtime	PropertiesUtils.resolveEnvPlaceholders	Re-interprets ${ENV_VAR[:default]} for all values in the loaded Properties

In other words, any ${...} remaining after deployment can still be filled by the process's environment variables. This is why Docker -e VAR=... injection takes precedence over template defaults.

PP_* Variables Not in Sidecar (Template defaults apply)

The following variables are referenced by the template as ${PP_XXX:기본값} but are not defined in the sidecar file. Therefore, the template default is always used unless explicitly injected.

PP_EXTERNAL_DB_PASSWORD · PP_MAIL_SMTP_HOST · PP_MAIL_SMTP_PORT · PP_MAIL_SMTP_AUTH · PP_MAIL_SMTP_USER · PP_MAIL_SMTP_PASSWORD · PP_MAIL_SMTP_STARTTLS · PP_DIAGNOSTIC_EMAIL

PP_AI_* was on this list but now does nothing

Providing PP_AI_OPENAI_*, PP_AI_MCP_*, or PP_AI_CHAT_ENABLED will be silently ignored. The layer that read those values was removed on 2026-09-18 — see AI Configuration.

PP_STORAGE_PASSWORD · PP_METASTORE_PASSWORD were on this list but are «non-existent variables»

Neither the template nor the sidecar references those names. The actual names are as follows:

Item	Actual Variable
Time series storage (Aster TSDB) connection	PP_CASSANDRA_HOST · PP_CASSANDRA_PORT · PP_CASSANDRA_USER · 🔑 PP_CASSANDRA_PASSWORD
Metastore (PostgreSQL) connection	PP_POSTGRES_HOST · PP_POSTGRES_PORT · PP_PG_USER · 🔑 PP_PG_PASSWORD

🔴 PP_MAIL_SMTP_* also now does nothing. The authoritative source for mail settings is pp.mm_config, and values are changed via the console (/config/index) or that table — Mail Configuration. This migration was made to ensure that values that must match as a set (host, port, authentication, account) are modified in one place only.

5 Types of Server Management Settings

There are 5 configuration files defined by the ConfigTemplates.SERVER_CONFIGS array and deployed as plantpulse-server/config/.

#	Template	Deployment Path
1	plantpulse-engine.properties	/plantpulse-server/config/plantpulse-engine.properties
2	plantpulse-storage.properties	/plantpulse-server/config/plantpulse-storage.properties
3	plantpulse-mq.properties	/plantpulse-server/config/plantpulse-mq.properties

These three files are read all at once by PropertiesLoader.load() (engine/mq/storage), and typed configuration beans EngineConfig · StorageConfig · MqConfig are reconstructed.

Mail and AI settings are not files — pp.mm_config is the authoritative source

plantpulse-mail.properties and plantpulse-ai.properties were deleted on 2026-09-17, and the path that read old copies remaining in the external configuration directory and moved them to a table was completely removed on 2026-09-18. In other words, even if you create those files, nothing will be read.

The only paths for values are the console and PostgreSQL pp.mm_config. The authoritative source for key names and defaults is in the code's MailConfigKeys · AiConfigKeys; keys without a row in the table are read using those defaults — the key tables in the two sections below remain valid.

application.properties (Removed)

application.properties was removed in June 2026. There is no code that reads this file (0 usages across all repos), and existing items were migrated as follows:

• Console behavior settings such as theme / homepage → Console System > Configuration Management (PostgreSQL mm_config table)
• alarm.duplicate.check.minutes → plantpulse-engine.properties series (read by EngineConfig)

---

plantpulse-engine.properties

Path: plantpulse-server/config/plantpulse-engine.properties · Template: template/plantpulse-engine.properties

Controls the core operation of the data processing engine.

Async / Pipeline

Property	Default	Substitution	Description (Template Comment)
engine.async.parallelism	256	—	Async worker pool parallelism (total cores ~= 256)
engine.pipeline.threads	36	—	Number of pipeline threads
engine.pipeline.ratelimit	40000	—	Pipeline processing limit per second (rate limit)
engine.pipeline.queue.size	1200000	—	Pipeline queue size (backpressure limit)
engine.pipeline.queue.o3.delay	50	—	Out-of-order data allowed delay (ms)
engine.pipeline.task.mode	SINGLE	—	Task execution mode (SINGLE / PARALLEL)

NTP

Property	Default	Substitution	Description
engine.ntpdate.server.ip	127.0.0.1	—	NTP server IP
engine.ntpdate.sync	false	—	Whether to perform NTP synchronization on boot

Stream / CEP

Property	Default	Substitution	Description
engine.stream.processor	DIRECT	—	Stream processing mode (DIRECT/BUFFERED)
engine.cep.server.ip	127.0.0.1	${PP_CEP_HOST:127.0.0.1}	CEP server IP
engine.cep.server.port	7400	${PP_CEP_CONNECT_PORT:7400}	CEP server port (auto-align by protocol: http→7400, https→7401)
engine.cep.server.protocol	http	${PP_CEP_PROTOCOL:http}	CEP protocol. Default http for internal communication, https for distributed/containers
engine.cep.server.api.key	(None)	${PP_CEP_API_KEY}	For CEPClient X-API-Key header. Must match CEP server cep.api.key

Streaming Message Delivery

Property	Default	Substitution	Description
engine.streaming.messaging.warning.ms	5000	—	Message delivery delay warning (ms)
engine.streaming.messaging.timeout.ms	10000	—	Message delivery timeout (ms)
engine.streaming.messaging.timeout.store.type	FILE_QUEUE	—	Timeout message storage method (FILE_QUEUE/MEMORY)
engine.streaming.messaging.timeout.recovery.type	DB	—	Timeout message recovery source (DB/FILE)

Job Worker Thread

Property	Default	Substitution	Description
engine.job.thread.asset	8	—	Number of asset job threads
engine.job.thread.point	8	—	Number of point job threads

DDS

Property	Default	Substitution	Description
engine.dds.enabled	true	—	Global DDS activation
engine.dds.data.tag.enabled	true	—	Tag data DDS publishing
engine.dds.data.asset.enabled	true	—	Asset data DDS publishing

Anomaly Detection / Data Flow

Property	Default	Substitution	Description
engine.data.anomaly.enabled	false	—	Anomaly detection activation (external service integration)
engine.data.anomaly.url	https://127.0.0.1:8970	—	Anomaly detection service URL
engine.dataflow.class	plantpulse.core.engine.pipeline.dataflow.BaseDataFlow	—	Data flow implementation class
engine.dataflow.server-timestamp.override	false	—	Whether to force server timestamp (true: collection time, false: original)
engine.dataflow.negative-timestamp.override	true	—	Whether to correct negative timestamps
engine.factory.model.updater.class	plantpulse.core.engine.model.impl.DefaultFactoryMetaModelUpdater	—	Factory metamodel updater
engine.event.handler.class	plantpulse.core.engine.event.DefaultEventHanlder	—	Event handler implementation (typo in class name Hanlder is as per actual class name)
engine.javagc.enabled	false	—	Whether to allow explicit System.gc() calls

Flow Engine

Property	Default	Substitution	Description
flow.debug.execution.log	false	—	Node execution INFO log. true for debugging sessions only
flow.stats.prune.interval.ms	86400000	—	Counter cleanup execution cycle (ms, 24 hours). Aster TSDB counter tables cannot use default_time_to_live, so cleanup is at the application level
flow.stats.prune.hourly.days	30	—	Retention period (days) of hour_bucket in tm_flow_stats_hourly
flow.stats.prune.daily.days	90	—	Retention period (days) of day_bucket in tm_flow_stats_daily
flow.health.check.interval.ms	600000	—	Flow health check cycle (ms, 10 minutes). ERROR if current time bucket errors > 0, otherwise NORMAL (KST based)
flow.webhook.api.key	dev-webhook-19ba7a54-1132-49f2-967e-3c6f5cd989a0	${PP_FLOW_WEBHOOK_API_KEY:...}	Flow engine external webhook authentication

Aster TSDB tables used by Flow health check: tm_flow_health_status (flow_id, hour_bucket — current time accumulator), tm_flow_health_status_10m (flow_id, min10_bucket — 10-minute time series).

Edge Health Check

Property	Default	Substitution	Description
edge.health.check.interval.ms	600000	—	Edge health check cycle (ms, 10 minutes)

The determination source is the _ON / _NORMAL flags in the Aster TSDB tm_edge_status payload. There is no RT REST call, so the load is zero and it completes within 1 second. Tables: tm_edge_health_status (edge_id, hour_bucket), tm_edge_health_status_10m (edge_id, min10_bucket).

Authentication / Security

Property	Default	Substitution	Description
api.key	(None)	${PP_API_KEY}	Platform internal API key (Server ↔ Edge Agent authentication)
engine.api.v5.auth.bruteforce.enabled	true	—	V5 API authentication brute force defense
engine.api.v5.auth.bruteforce.ip_limit	30	—	Attempt limit per IP
engine.api.v5.auth.bruteforce.user_limit	10	—	Attempt limit per user
engine.api.v5.auth.bruteforce.window.seconds	60	—	Counting window (seconds)
engine.api.v5.trusted_proxies	(Empty)	—	Trusted proxy IP whitelist. Only when configured, if remoteAddr is on the list, the first IP of X-Forwarded-For is used as client IP. If empty, only getRemoteAddr() is used (fail-safe)
engine.session.bruteforce.enabled	true	—	/login/login browser session brute force defense
engine.session.bruteforce.ip_limit	30	—	Attempt limit per IP
engine.session.bruteforce.user_limit	10	—	Attempt limit per user
engine.session.bruteforce.window.seconds	300	—	Counting window (seconds)

The session brute force counter is Caffeine in-memory — multi-node sharing (Redis) is not implemented.

API Token

Property	Default	Substitution	Description
engine.api.token.cleanup.enabled	true	—	Expiry(expires_at < now)/Revocation(revoked_at IS NOT NULL) token cleanup
engine.api.token.cleanup.retention_days	30	—	Hard delete from mm_token after this period expires. Once every 24 hours (first run 60s after boot)
engine.api.token.default_ttl_days	90	—	Default expiry (days) when issuing new tokens. 0 = permanent token. If the issuance request provides ttl_days, that value takes precedence → Issue API Token

Existing tokens where expires_at is NULL are maintained as permanently active (backward compatibility).

---

plantpulse-storage.properties

Path: plantpulse-server/config/plantpulse-storage.properties · Template: template/plantpulse-storage.properties

Manages connections and storage policies for Metastore (PostgreSQL), Cache (Valkey), Time Series (Aster TSDB), TSE, and Data Gateway.

⚠️ Typo keys actually exist in this file. Three keys have spellings different from what the code reads — you must read the Typo Key Warning section.

PostgreSQL Metastore

Property	Default	Substitution	Description
metastore.driver	org.postgresql.Driver	—	JDBC driver class
metastore.url	jdbc:postgresql://...	${PP_POSTGRES_HOST} · ${PP_POSTGRES_PORT} · ${PP_DB_NAME}	Metastore JDBC URL. Includes query string ?characterEncoding=UTF-8&reWriteBatchedInserts=true
metastore.user	—	${PP_PG_USER}	Metastore account
metastore.password	—	${PP_PG_PASSWORD}	Metastore password

Cache (Valkey / Redis Compatible)

Property	Default	Substitution	Description
cache.host	—	${PP_REDIS_HOST}	Cache host
cache.port	—	${PP_REDIS_PORT}	Cache port (internal plaintext Valkey backend)
cache.password	—	${PP_REDIS_PASSWORD}	Cache password
cache.ssl.enabled	false	—	Cache TLS. Internal clients default to redis://
cache.ssl.truststore.location	/var/security/plantpulse/master/master.truststore.p12	—	Truststore path
cache.ssl.truststore.password	kopens123!	${PP_TLS_TRUSTSTORE_PASSWORD:kopens123!}	Truststore password
cache.ssl.truststore.type	PKCS12	—	Truststore type
cache.ssl.endpoint.identification.enabled	false	—	Endpoint identity verification

The authoritative source for key constants is InMemoryConfigKeys in plantpulse-inmemory.

TSE (TimeSeries Engine)

Property	Default	Substitution	Description
tse.protocol	http	${PP_TSE_PROTOCOL:http}	TSE protocol. Default plaintext http for internal communication (2026-08)
tse.host	—	${PP_TSE_HOST}	TSE host
tse.port	7800	${PP_TSE_CONNECT_PORT:7800}	TSE port. Default plaintext listener 7800
tse.user	tse	${PP_TSE_USER:tse}	TSE account
tse.password	tse123!	${PP_TSE_PASSWORD:tse123!}	TSE password

Aster TSDB Time Series Storage

Property	Default	Substitution	Description
storage.db_type	CASSANDRA	—	DB type. CASSANDRA/DSE/SCYLLADB remain in template comments, but the only currently supported value is CASSANDRA
storage.db_version	5.0	—	Not the server version, but «feature compatibility level» — see note below
storage.host	—	${PP_CASSANDRA_HOST}	Contact point host

If storage.db_version is 5.0, but nodetool version appears as 6.0.0

Both are correct. They are not values referring to the same thing.

	Value	What
storage.db_version	5.0	Compatibility level determining which feature set the driver speaks with
Actual Bundled Server	6.0.0	apache-cassandra-timeseries-6.0.0.jar — Internal fork of cassandra-6.0

Do not read 5.0 and assume «the server is 5.0». Information regarding the server is in Time Series Database. Do not arbitrarily increase this value to 6.0 — driver behavior will change.

| storage.port | 9042 | ${PP_STORAGE_PORT:9042} | Native protocol port | | storage.keyspace | — | ${PP_KEYSPACE} | Keyspace used | | storage.user | — | ${PP_CASSANDRA_USER} | Connection account | | storage.password | cassandra | ${PP_STORAGE_PASSWORD:cassandra} | Password. PP_STORAGE_PASSWORD is not exported in env.sh → default applied | | storage.durable_writes | true | — | Use commit log (⚠️ No reading code — see below) | | storage.append_columns | [] | — | Auto-add user-defined columns | | storage.replication | { 'class' : 'NetworkTopologyStrategy', 'datacenter1' : 1 } | — | Replication strategy (Single node default) | | storage.data.disk.name | — | ${PP_DATA_DISK_NAME} | Data disk name |

TTL (Unit: Days)

Property	Default	Description
storage.tag.point.ttl	62	Tag time series original
storage.tag.point.map.ttl	1	Tag point map
storage.tag.point.sampling.ttl	93	Tag sampling
storage.tag.point.snapshot.ttl	93	Tag snapshot
storage.tag.point.aggregation.ttl	93	Tag aggregation
storage.tag.point.archive.ttl	365	Tag archive (long-term retention)
storage.tag.blob.ttl	93	Tag BLOB
storage.asset.data.ttl	10	Asset data
storage.asset.data.sampling.ttl	31	Asset sampling

Asset Snapshot Cycle

Property	Default	Description
storage.asset.data.snapshot.interval	10 SECONDS	Records the tag map (data_map) of assets before each cycle to the asset_data series (data/second/timestamp + min/hour/day boundaries)

Template comments explain the reasoning: 1 SECONDS was congested with approx 22K writes per minute based on 121 assets (approx 4% of total Aster TSDB). Since tag originals (second unit) are fully preserved in tm_tag_point/TSE, asset view snapshots were lowered to 10s (matching code default) — reducing writes by 90% while maintaining record integrity.

This value becomes the ASSET_DATA_JOB cycle of the scheduler. It must be in "<n> <unit>" 2-token format; refer to the scheduler architecture for rules.

Internal Metastore Synchronization (Experimental)

Property	Default	Description
storage.internal.metastore.enabled	false	Internal metastore synchronization activation
storage.internal.metastore.thread.count	4	Number of synchronization worker threads
storage.internal.metastore.queue.size	12000000	Synchronization queue size
storage.internal.metastore.point.retation.days	10	Point retention period (⚠️ Typo key — see below)

Row Cache

Property	Default	Description
storage.asset.data.row.cache.size	NONE	Asset data row cache (NONE/ROWS_ONLY/ALL) — ⚠️ No reading code
storage.tag.point.row.cache.size	NONE	Tag point row cache — ⚠️ No reading code

Compaction

Property	Default	Description
storage.table.compaction.strategy	UCS	Default strategy (TWCS: TimeWindow, UCS: Unified). Read by CassandraCreateDAO (Code fallback: TWCS)
storage.table.compaction.strategy.twcs.window_size	1	TWCS window size — ⚠️ No reading code
storage.table.compaction.strategy.twcs.window_unit	DAYS	TWCS window unit — ⚠️ No reading code
storage.table.compaction.strategy.ucs.scailing_parameter	T8	UCS scaling parameter (T8 = tiered 8 levels). Key spelling scailing matches code
storage.table.compaction.strategy.ucs.min_sstable_size	128MiB	UCS minimum sstable size (Code fallback: 100MiB)
storage.table.compaction.strategy.ucs.target_sstable_size	512MiB	UCS target sstable size
storage.table.compaction.strategy.ucs.base_shard_count	8	UCS base shard count
storage.table.compaction.strategy.ucs.max_sstables_to_compact	6	Maximum number of sstables to compact at once
storage.table.compaction.strategy.ucs.sstable_growth	0.7	UCS sstable growth ratio

Compression / Query Limits / Backup / Data Gateway

Property	Default	Substitution	Description
storage.compresion.zstd.level	3	—	zstd compression level (1~22) — ⚠️ Typo key, see below
storage.tag.point.select.limit.size	1000000	—	Max rows for single tag point select
storage.doamin.model.changed.to.metastore.backup	true	—	Metastore backup upon domain model change — ⚠️ Typo key + No reading code
data.gateway.protocol	http	${PP_DATA_GATEWAY_PROTOCOL:http}	Data Gateway protocol
data.gateway.host	127.0.0.1	${PP_DATA_GATEWAY_HOST:127.0.0.1}	Data Gateway host (assumes local co-location)
data.gateway.port	5500	${PP_DATA_GATEWAY_CONNECT_PORT:5500}	Data Gateway port (Code fallback: 5501)
data.gateway.api.key	—	${PP_DATA_GATEWAY_API_KEY}	Data Gateway API key

⚠️ Typo Key — Ignored even if edited

The following 3 keys actually existing in the template have spellings different from what the code reads. Even if an administrator changes the values in these lines, they will be silently ignored, and the code defaults will be applied.

Key in Template (Typo)	Key actually read by code	Code Default	Read Location
storage.compresion.zstd.level	storage.compression.zstd.level	3	StorageConfig → CassandraCreateDAO
storage.internal.metastore.point.retation.days	storage.internal.metastore.point.retention.days	10	StorageConfig → PointDAO
storage.doamin.model.changed.to.metastore.backup	None (No code reads this even with correct spelling)	—	—

To actually change the value, add the key with the correct code spelling to the same file. For example, to increase the zstd level to 6, you must add storage.compression.zstd.level=6 (do not just fix the typo key, but add a new line with the correct spelling). Leaving the typo line is harmless, but fixing the template itself is the fundamental solution.

No code reads storage.doamin.model.changed.to.metastore.backup regardless of spelling. Backup during domain model change is handled by MetastoreBackupDeployer (cron 0 0 12 1 1/1 ? *, 1st of every month 12:00), not this key — refer to the scheduler architecture.

Storage Keys Read by Code but Missing from Template

Property	Code Default	Description
storage.compression.zstd.type	ZstdCompressor	Compressor class name

Storage Template Keys with No Reading Code (Invalid if edited)

storage.durable_writes · storage.asset.data.row.cache.size · storage.tag.point.row.cache.size · storage.table.compaction.strategy.twcs.window_size · storage.table.compaction.strategy.twcs.window_unit (+ the 3 typo keys above)

Even if the two keys related to TWCS are changed to storage.table.compaction.strategy=TWCS, they cannot control window size/unit.

---

plantpulse-mq.properties

Path: plantpulse-server/config/plantpulse-mq.properties · Template: template/plantpulse-mq.properties

Manages Kafka / MQTT broker connections and DDS publishing topics. The code-side SSOTs are MqConfig (connections) and UNSTopicDefaults (topics).

Broker Connection

Property	Default	Substitution	Description
mq.host	—	${PP_KAFKA_HOST}	Broker host (Shared by Kafka/MQTT)
mq.user	—	${PP_MQ_USER}	Broker admin account
mq.password	—	${PP_MQ_PASSWORD}	Broker password
mq.kafka.port	9092	${PP_KAFKA_PORT:9092}	Kafka port
mq.kafka.security.protocol	SASL_PLAINTEXT	—	Kafka security protocol
mq.ssl.truststore.location	/var/security/plantpulse/master/master.truststore.p12	—	Messaging TLS truststore (MQTT + optional Kafka TLS)
mq.ssl.truststore.password	kopens123!	${PP_TLS_TRUSTSTORE_PASSWORD:kopens123!}	Truststore password
mq.ssl.truststore.type	PKCS12	—	Truststore type
mq.ssl.endpoint.identification.algorithm	(Empty)	—	Hostname verification algorithm (Empty = No verification)
mq.mqtt.port	—	${PP_MQTT_TLS_PORT}	MQTT connection port
mq.mqtt.ssl.enabled	true	—	MQTT TLS use

In addition to the keys above, MqConfig also reads Kafka-specific truststore overrides mq.kafka.ssl.truststore.location / .password / .type and mq.kafka.ssl.endpoint.identification.algorithm (not in template, defaults to mq.ssl.* if absent).

DDS Topics

14 Kafka types (mq.dds.kafka.topic.*) + 14 MQTT UNS types (mq.dds.mqtt.topic.*) + 2 SparkplugB types (mq.dds.mqtt.sparkplugb.*).

The complete list of keys and defaults is organized in Messaging Architecture → DDS Publishing Topics. Topic values are Kafka/MQTT wire contracts, and UNSTopicDefaultsContractTest freezes the literals, so arbitrary changes are prohibited.

${PP_TOPIC_PREFIX} simultaneously determines the prefix of Kafka topics (pp-), the root of MQTT UNS paths (pp/), and the SparkplugB group ID.

---

Mail Configuration (pp.mm_config)

Authoritative source: PostgreSQL pp.mm_config · Key authoritative source: plantpulse.core.config.MailConfigKeys. Reading side: SMTPMailSender.loadMailConfig().

🔴 PP_MAIL_* environment variables also do nothing

Same as AI — the substitution layers for plantpulse-mail.properties and ${PP_MAIL_*} were removed on 2026-09-17~18. Providing them via sidecar or docker run -e will be silently ignored. Enter values in the console Configuration Screen(/config/index) or the pp.mm_config table.

Config Key	Default	Description
MAIL_SEND_ENABLED	false	Master Sending Gate. If off, skip without SMTP connection and leave only WARN
MAIL_SMTP_HOST	smtp-relay.brevo.com	SMTP relay address
MAIL_SMTP_PORT	587	587 = STARTTLS, 465 = Implicit TLS (with MAIL_SMTP_SSL_ENABLE)
MAIL_SMTP_AUTH	true	Whether to use SMTP login. Even if on, if MAIL_SMTP_USER is empty, no login is performed
MAIL_SMTP_USER	(Empty)	SMTP account. May differ from sender address as it is the name issued by the relay. Can be saved as empty
🔑 MAIL_SMTP_PASSWORD	(Empty)	SMTP password or API key. Authentication will be rejected if empty on a login-requiring relay
MAIL_SMTP_STARTTLS_ENABLE	true	Whether to upgrade to STARTTLS on 587
MAIL_SMTP_SSL_ENABLE	false	Whether to connect via TLS from the start like 465. In this case, starttls is disabled
MAIL_SMTP_CONNECTIONTIMEOUT	10000	Connection timeout (ms). If absent, JavaMail default is «infinite wait»
MAIL_SMTP_TIMEOUT	10000	Read timeout (ms). 〃
MAIL_SMTP_WRITETIMEOUT	10000	Write timeout (ms). 〃
MAIL_FROM	(Empty)	Sender address. If empty, MAIL_SMTP_USER is used, but relay account names are often not valid addresses and may be rejected by the recipient — use a real address
sms.sender.class	plantpulse.core.service.support.alarm.sms.DefaultSMSSender	SMS sender implementation FQN. If a non-existent class is entered, it reverts to default implementation

TLS only uses 1.3 · 1.2 and verifies server certificate names.

SMTP credentials are no longer in defaults

The defaults for MAIL_SMTP_USER · MAIL_SMTP_PASSWORD are empty values. Since MAIL_SEND_ENABLED is the default false, sending will not be attempted until an operator enables it.

Sending via Gmail SMTP Relay

Google Workspace SMTP relay (smtp-relay.gmail.com) can be authorized by the sending server's IP address. In this case, since there is no account, leave login disabled. First, register the outgoing IP of this server in the Google Admin Console's SMTP relay settings.

Config Key	Value
MAIL_SEND_ENABLED	true
MAIL_SMTP_HOST	smtp-relay.gmail.com
MAIL_SMTP_PORT	587
MAIL_SMTP_AUTH	false
MAIL_SMTP_USER	(Empty)
MAIL_SMTP_STARTTLS_ENABLE	true
MAIL_SMTP_SSL_ENABLE	false
MAIL_FROM	Actual address of Workspace domain

After saving, click Re-apply on the configuration screen.

Alarm Mail

• The product logo is included in the mail header (CID inline image). It is visible even without a console address and is not fetched externally.
• The location line of the mail is written starting from the site (Site › Area › Line › Equipment; company is excluded).
• The «Go to check alarm» link is PP_UI_CONSOLE_URL (<콘솔 주소>/alarm/view/<알람 번호>). If empty, the mail is sent as is, but only the link is omitted — Address Configuration.

diagnostic.log.email is gone

It was removed from the table as there is no code that reads it (product testing prevents that key from returning). Even if it remains somewhere, it does nothing.

---

AI Configuration (pp.mm_config)

Authoritative source: PostgreSQL pp.mm_config · Key authoritative source: plantpulse.core.config.AiConfigKeys. Reading side: AiConfig (+ AiConfig.mcpSettings()).

🔴 PP_AI_* environment variables do nothing

Old plantpulse-ai.properties was deleted on 2026-09-17, and the path that read that file and the ${PP_AI_*} substitution layer were both removed on 2026-09-18. Providing names like PP_AI_OPENAI_ENABLED via sidecar or docker run -e will be silently ignored — no errors or warnings.

This is why the product removed those names instead of keeping them «just in case». Names that exist but are ignored are worse than non-existent names because the server appears to be configured.

There are only two ways to enter values — the configuration screen in the console (/env), or directly modifying the pp.mm_config table. Keys without a row are read as the defaults in the table below; a row is created when a value is entered.

LLM Gateway

Property	Default	Description
ai.llm.enabled	false	Global AI switch. If off, all AI screens and endpoints are closed
ai.llm.base.url	https://api.openai.com	OpenAI compatible gateway (LiteLLM/vLLM). Do not append / to the end
ai.llm.chat.path	/v1/chat/completions	Change only when the gateway uses a different path
ai.llm.responses.path	/v1/responses	〃
🔑 ai.llm.api.key	(Empty)	Gateway key. Calls will be rejected with 401 if empty
ai.llm.model	gpt-5-mini	Model name exposed by the gateway. If base.url is changed, change this as well
ai.llm.timeout.ms	60000	LLM call timeout (ms)
ai.llm.max.output.tokens	4000	Maximum response tokens
ai.llm.temperature	-1	If negative, omit entirely from the request to use the model's default

Old name ai.openai.* is no longer read

The prefix has changed to ai.llm.. Even if a ai.openai.base.url row remains in pp.mm_config, the code doesn't read it and uses the default (https://api.openai.com) — this creates a situation where it looks like the gateway is configured, but it attempts to go to OpenAI and fails. Check these rows on upgraded installations.

Why is ai.llm.temperature negative by default

The gpt-5 series does not support temperature and returns 400 Unsupported value if a value other than the default 1 is sent. Therefore, a negative value meaning «do not send» is used as the default. Increase it to 0 or above only when using other models behind LiteLLM/vLLM.

Feature Switches

🔴 Defaults changed on 2026-09-16. The only AI feature remaining on the screen is Chat; others were removed by the product. Returning values to true will bring back the features, but those screens are not in a validated state for this version.

Property	Default	Description
ai.chat.enabled	true	AI Chat Console — This is the only AI feature remaining on the screen
ai.chat.history.ttl.days	90	Chat history retention (days). Applied as Cassandra TTL and reflected from «new» messages after restart
ai.alarm.rca.enabled	false	Alarm RCA — if on, AI root cause analysis is added to alarm details
ai.alarm.eql.enabled	false	Alarm EQL Assistant — if on, a draft button is added to the alarm configuration form
ai.uns.topic.enabled	false	UNS Topic Description — if on, a description panel is added to the UNS tree
ai.result.audit.enabled	false	AI Result Audit — Internal RCA component without a screen. Enable this when enabling RCA
ai.similar.incident.enabled	false	Similar Incident Evidence Collection — 〃
ai.similar.incident.max.items	5	Maximum number of similar incidents to include as evidence
ai.similar.incident.min.score	40	Minimum score to be recognized as a similar incident

6 keys removed in 2026-08

3 keys for ai.diagnostic.summary.* (AI summary of diagnosis logs) and 3 keys for ai.operations.report.* (AI operation report) were removed from the product on 2026-08-25. Screens, JS, and controllers were also removed, so the features will not return even if values are restored. Leaving them causes no harm — no code reads them.

MCP

🔴 MCP is «independent» from ai.llm.enabled. Since it is for external clients, not screens, /api/v5/mcp remains open even if the gateway is off. For details, see MCP Server.

Property	Default	Description
AI_MCP_ENABLED	true	If false, then /api/v5/mcp returns 404
AI_MCP_ALLOWED_ORIGINS	(Empty)	Browser Origin whitelist (comma separated). If empty, all browser requests are rejected, allowing only native clients
AI_MCP_TREE_MAX_NODES	2000	Tree response node upper limit — dependent on model context window; increase as model size increases
AI_MCP_TREE_DEFAULT_DEPTH	2	Tree query default depth
AI_MCP_TREE_MAX_NODE_TAGS	100	Tag count upper limit per node
AI_MCP_TREE_ONTOLOGY_MAX_CHARS	200000	Ontology response character upper limit
AI_MCP_TREE_ONTOLOGY_DEFAULT_DEPTH	3	One level deeper than default tree depth — must reach equipment nodes with tag counts to answer «how many tags for this equipment» as a default response
AI_MCP_TREE_ONTOLOGY_MAX_TAGS	500	Tag row budget for the entire ontology response (separate from per-node limit). Since one slim tag is approx 85B, loading the entire site consumes 33~56% of the context window

---

Other Templates

template/ contains application and infrastructure configuration templates in addition to the 5 types above. The authoritative deployment path mapping is ConfigTemplates.java.

Template	Deployment Path	Purpose
plantpulse-batch.properties	/plantpulse-batch/config/	Batch (Time series sync/aggregation)
plantpulse-cep.properties	/plantpulse-cep/config/	CEP Engine
plantpulse-jdbc.properties	/plantpulse-data-gateway/config/	JDBC Integrated Configuration
plantpulse-data-gateway.properties	/plantpulse-data-gateway/config/	Data Gateway Operation
plantpulse-warehouse.properties	/plantpulse-warehouse/config/	Aster TSDB → Iceberg(S3) Archive
plantpulse-datalake-admin-api.properties	/plantpulse-datalake-admin-api/config/	Admin Console API · JMX Collection
plantpulse-plugin-opcua.properties	/plantpulse-plugin/opc-ua/config/	OPC UA Server Plugin
plantpulse-plugin-aas.properties	/plantpulse-plugin/aasx-server/config/	AAS V3 (aasx-server) Plugin
kafka.properties	/plantpulse-messaging/kafka/config/kafka.properties	Kafka(KRaft) Broker
kafka-jaas.conf	/plantpulse-messaging/kafka/config/jaas.conf	Kafka JAAS
hivemq.xml	/plantpulse-messaging/mqtt/conf/config.xml	HiveMQ Broker
plantpulse-mq-auth.properties	/plantpulse-messaging/mqtt/conf/auth.properties	HiveMQ Auth Source
hive-auth.properties	/plantpulse-analytics/spark/conf/	Hive/Analysis Account Auth Source
plantpulse-timeseries-engine.conf	/plantpulse-timeseries/engine/conf/ (2 locations)	TSE
valkey.conf · postgresql.conf · pg_hba.conf · cassandra.yaml · jvm-server.options	/plantpulse-storage/...	Storage
spark-env.sh · spark-defaults.conf · hive-site.xml · kyuubi-defaults.conf · gravitino-iceberg-rest-server.conf	/plantpulse-analytics/...	Analysis
workflow.yaml · application.yaml	/plantpulse-workflow/...	Temporal / Kestra

plantpulse-websocket.properties also does not exist. There are no WebSocket/STOMP series configuration files. template/cluster/ is a worker-node-specific override — if the master template is modified, this must also be synchronized.

plantpulse-batch.properties

Consists of connection blocks (Platform/Metastore/Cache/TSE/Kafka/Aster TSDB/Spark) and batch operation settings. Note that connection keys differ from storage/jdbc — e.g., Aster TSDB uses storage.db.host / storage.db.port / storage.db.keyspace (server uses storage.host / storage.port / storage.keyspace).

Property	Default	Description
platform.host	${PP_MASTER_IP}	Platform(server-web) host
mq.port	9092	Kafka port (Literal — not substituted)
kafka.topic	pp-batch	Batch input topic
kafka.group.id	PP-BATCH-GROUP	Batch consumer group
batch.timer_schedule_delay	5000	Batch timer cycle (ms)
batch.thread_pool_batch_thread_size	4	Number of batch worker threads
batch.buffer_drain_limit_size	10000	Max items to empty at once
batch.sync.inmemory.enabled	false	In-memory synchronization
batch.sync.inmemory.ttl.minute	10	In-memory synchronization TTL (minutes)
batch.sync.tse.enabled	true	Distribution to TSE
batch.sync.external.db.enabled	false	External DB synchronization
external.db.name	TIMESCALEDB	External DB type. INFLUXDB/QUESTDB blocks are in comment state in the template
analytics.url	jdbc:spark://${PP_SPARK_HOST}:${PP_KYUUBI_PORT}/default	Spark Thrift JDBC

plantpulse-cep.properties

Property	Default	Description
cep.backend	aster	Backend to boot (aster · flink). Any other value causes boot failure — Backends
cep.aster.state.enabled	true	Preserve Aster CEP rule states (window · aggregation · in-progress pattern) in cache.* Redis. Key is PP:CEP:ASTER-STATE:<문장>
cep.aster.state.snapshot_interval_ms	60000	Snapshot cycle — maximum state that can be lost during crash. Normal shutdown takes a final snapshot before exiting
cep.async.core_pool_size	32	Async core pool
cep.async.max_pool_size	256	Async max pool
cep.async.queue_capacity	50000	Async queue
cep.async.keep_alive_seconds	60	Thread keep-alive
cep.async.shutdown_timeout_seconds	30	Graceful shutdown wait
cep.consumer.thread_count	16	Consumer thread
cep.stream.shard_count	16	Stream shard count
cep.stream.drain_limit_size	1000	Drain limit
cep.stream.drain_warn_ms	1000	Drain warning (ms)
cep.monitor.initial_delay_seconds	5	Monitor initial delay
cep.monitor.interval_seconds	1	Monitor cycle
cep.monitor.store_interval_seconds	60	Aster TSDB metric storage cycle
cep.api.key	${PP_CEP_API_KEY}	EQL management endpoint authentication (X-API-Key)
cep.api.max_eql_text_length	100000	Max EQL text length (DoS defense)
cep.storage.request_timeout_seconds	10	Aster TSDB request timeout
cep.storage.connect_max_retries	3	Connection retry
cep.storage.ttl_seconds	864000	Metric TTL (seconds, 10 days)
cep.stream.pending_idle_threshold_ms	60000	Pending idle threshold
cep.stream.cleaner_initial_delay_ms	60000	Cleaner initial delay
cep.stream.cleaner_interval_ms	10000	Cleaner cycle
cep.stream.max_clean_duration_ms	30000	Max clean duration
cep.stream.cleaner_max_claim_count	5000	Max claim count
cep.stream.ack_chunk_size	500	ACK chunk size
cep.stream.max_deliveries	3	Events are sent to dead letter only after being delivered this many times without being processed
cep.stream.dead_letter.retention_ms	86400000	Dead letter retention age (24 hours)
cep.stream.dead_letter.max_bytes	268435456	Dead letter byte budget (256 MiB, total across shards). If exceeded, oldest entries are cut from each shard

plantpulse-jdbc.properties

Integrated connection settings for external JDBC clients (SQL module · Data Gateway). Prefix is jdbc.<tech> (postgres/cassandra/spark/tse).

Property	Value	Description
jdbc.postgres.driver	org.postgresql.Driver	META / PQL
jdbc.postgres.url	jdbc:postgresql://${PP_POSTGRES_HOST}:${PP_POSTGRES_PORT}/${PP_DB_NAME}?characterEncoding=UTF-8
jdbc.cassandra.driver	plantpulse.cassandra.jdbc.CassandraDriver	OLTP Time Series(CQL). Driver value is for configuration consistency (not used functionally)
jdbc.cassandra.url	jdbc:cassandra://${PP_CASSANDRA_HOST}:${PP_CASSANDRA_PORT}/${PP_SCHEME}
jdbc.tse.driver	plantpulse.timeseries.tql.TQLDriver	OTAP / TQL
jdbc.tse.url	jdbc:ts://${PP_TSE_HOST}:${PP_TSE_PORT}/${PP_SCHEME}	⚠️ Fixed to plain http(ts/7800)
jdbc.spark.driver	plantpulse.spark.jdbc.SparkSQLDriver	OLAP Analysis
jdbc.spark.url	jdbc:spark://${PP_SPARK_HOST}:${PP_KYUUBI_PORT}/default/${PP_SCHEME}
data.gateway.api.key	${PP_DATA_GATEWAY_API_KEY}	For JDBC call authentication

⚠️ Do not change jdbc.tse.url to https(tss/7801). The TQL JDBC driver uses the plaintext TSE listener (7800). TSE clients default globally to plaintext http/7800 (2026-08, env.sh TSE section). If caught as tss/7801, JDBC connection fails → data-gateway boot fails → all /api/v1/* return 404 → analytics catalog 404. Therefore, it is pinned to ts/PP_TSE_PORT.

jdbc.spark.username is the analysis-only account ${PP_ANALYTICS_USER}. It is separate from the Kyuubi internal metastore account (PP_HIVE_*). Optional keys jdbc.<tech>.pool.initial|min|max (postgres/spark/tse) and jdbc.<tech>.query.timeout|fetch-size can be used (full reference in plantpulse-jdbc/docs/configuration.md). jdbc.cassandra.local-datacenter is in a comment state in the template and is datacenter1 when unconfigured.

plantpulse-data-gateway.properties

Property	Default	Description
data.gateway.api.key	${PP_DATA_GATEWAY_API_KEY:ADFA-URDV-QWED-1234}	API authentication key (Required). Dev default to be replaced upon delivery
data.gateway.rate.limit	1000	Max requests per second (Based on IP + API key)
data.gateway.slow.query.ms	10000	Slow query warning threshold (ms)
data.gateway.slow.update.ms	2000	Slow update warning threshold (ms)
data.gateway.async.timeout.ms	60000	Async timeout (ms)

plantpulse-datalake-admin-api.properties

Collection of JMX/management endpoints for each service based on a single node (type=MASTER). This is the only file that uses underscores (_) instead of dots (.) for keys.

Property	Default	Description
type	MASTER	Single node or cluster master
db_type	CASSANDRA	Monitored DB type
master / host	${PP_MASTER_IP} / ${PP_HOST_IP}	Master / Self host
db_host · db_port · db_keyspace · db_username · db_password · db_replication_factor	${PP_CASSANDRA_*} · 9042 · ${PP_SCHEME} · 1	Aster TSDB connection
tse_host · tse_port · tse_protocol · tse_username · tse_password	${PP_TSE_*} · 7801 · https · tse · tse123!	TSE connection (Metric loading)
kafka_jmx_uri	service:jmx:rmi:///jndi/rmi://${PP_KAFKA_HOST}:7299/jmxrmi	Kafka JMX
mqtt_jmx_uri	...://${PP_MQTT_HOST}:7279/jmxrmi	HiveMQ JMX
cep_jmx_uri	...://${PP_HOST_IP}:6499/jmxrmi	CEP JMX
datagateway_jmx_uri	...://${PP_HOST_IP}:7499/jmxrmi	Data Gateway JMX
sql_jmx_uri	...://${PP_HOST_IP}:7599/jmxrmi	SQL JMX
tse_jmx_uri	...://${PP_HOST_IP}:7899/jmxrmi	TSE JMX
cassandra_jmx_uri	...://${PP_CASSANDRA_HOST}:7199/jmxrmi	Aster TSDB JMX
spark_jmx_uri	...://${PP_SPARK_HOST}:10010/jmxrmi	Spark JMX
redis_jmx_uri	redis://${PP_REDIS_HOST}:${PP_REDIS_PORT}?password=...	Valkey connection URI
redis_ssl_truststore_location / _password / _type / redis_ssl_endpoint_identification_enabled	p12 · ${PP_TLS_TRUSTSTORE_PASSWORD:kopens123!} · PKCS12 · false	Valkey TLS
postgres_jmx_uri	jdbc:postgresql://.../${PP_DB_NAME}?user=...&password=...	PostgreSQL connection URI

JMX Port Summary: HiveMQ 7279, Kafka 7299, CEP 6499, Data Gateway 7499, SQL 7599, TSE 7899, Aster TSDB 7199, Spark 10010.

plantpulse-warehouse.properties

Aster TSDB → Iceberg(S3) archive batch. Only this file uses UPPER_SNAKE_CASE keys (UPPER_SNAKE).

Group	Key	Default
Temporal	TEMPORAL_SERVER · TEMPORAL_TASK_QUEUE · TEMPORAL_WORKFLOW_ID · TEMPORAL_SCHEDULE_ID · OPTIMIZER_SCHEDULE_ID · OPTIMIZER_WORKFLOW_ID	${PP_TEMPORAL_HOST}:${PP_TEMPORAL_PORT} · PP-S3-QUEUE · PP-S3-ARCHIVER-DAILY-WORKFLOW · PP-S3-ARCHIVER-DAILY · PP-S3-OPTIMIZER-WEEKLY · PP-S3-OPTIMIZER-WEEKLY-WORKFLOW
Schedule	DAILY_HOUR · DAILY_MINUTE · OPTIMIZER_DAY_OF_WEEK · OPTIMIZER_HOUR · OPTIMIZER_MINUTE · TIMEZONE	2 · 20 · 0(Sunday) · 3 · 0 · ${PP_TZ}
Execution Mode	ARCHIVER_MODE	SHELL (or KESTRA)
Kestra	KESTRA_SERVER · KESTRA_NAMESPACE · KESTRA_TASK_ID · KESTRA_USER · KESTRA_PASS	${PP_KESTRA_HOST}:${PP_KESTRA_PORT} · plantpulse · plantpulse-daily-s3-archiver · admin@plantpulse.io · (Empty)
Script	S3_SCRIPT_DIR · S3_SCRIPT_NAME · S3_OPTIMIZER_SCRIPT_NAME	${PP_HOME}/plantpulse-warehouse/s3 · archive.sh · optimize.sh
Spark	SPARK_APP_NAME · SPARK_MASTER · SPARK_JAR_PATH · SPARK_DRIVER_MEMORY · SPARK_EXECUTOR_MEMORY · SPARK_EXECUTOR_CORES · SPARK_NUM_EXECUTORS · ARCHIVER_PARALLELISM · SPARK_CHECKPOINT_DIR	PP-SPARK-S3-LAKEHOUSE · local[*] · .../plantpulse-warehouse-s3.jar · 2g · 4g · 4 · 3 · 4 · s3a://${PP_S3_BUCKET:-plantpulse}/spark-checkpoint/
Source/Target	CASSANDRA_CATALOG · CASSANDRA_DATABASE · S3_BUCKET · ICEBERG_CATALOG · ICEBERG_DATABASE	cassandra · ${PP_SCHEME:pp} · ${PP_S3_BUCKET:-plantpulse} · spark_catalog · ${PP_SCHEME:pp}
Optimization	SNAPSHOT_RETENTION_DAYS · OPTIMIZER_MIN_INPUT_FILES · OPTIMIZER_TARGET_FILE_SIZE_BYTES	3 · 10 · 134217728 (128MB)
Timeout	ARCHIVER_WF_EXECUTION/RUN/TASK_TIMEOUT_HOURS · OPTIMIZER_WF_EXECUTION/RUN/TASK_TIMEOUT_HOURS · ACTIVITY_MAX_ATTEMPTS · ACTIVITY_TIMEOUT_HOURS · OPTIMIZER_ACTIVITY_TIMEOUT_HOURS · SCRIPT_TIMEOUT_SECONDS · OPTIMIZER_TIMEOUT_SECONDS · HEARTBEAT_INTERVAL_SECONDS	12/12/12 · 8/7/12 · 3 · 12 · 8 · 21600(6h) · 25200(7h) · 60
Web	WEB_SERVER_PORT	9600

Messaging / Auth Source Template

plantpulse-mqtt.properties — HiveMQ broker configuration.

Property	Default
mqtt.host · mqtt.user · mqtt.password	${PP_MQTT_HOST} · ${PP_MQ_USER} · ${PP_MQ_PASSWORD}
mqtt.port · mqtt.ssl.enabled	${PP_MQTT_TLS_PORT} · true
mqtt.websocket_port	8888
mqtt.data_dir	/data1/pp-data/hivemq

plantpulse-mq-auth.properties — Auth source pointed to by HiveMQ security extension (FileAuthAuthenticator) via HIVEMQ_AUTH_FILE environment variable. Only two keys: mqtt.user / mqtt.password.

hive-auth.properties — Read by plantpulse-hive-auth (PasswordAuthenticator). Either HIVE_AUTH_FILE environment variable or $PP_HOME/plantpulse-analytics/spark/conf/hive-auth.properties. Notably, the key name itself is a placeholder.

${PP_HIVE_USER}=${PP_HIVE_PASSWORD}
${PP_ANALYTICS_USER}=${PP_ANALYTICS_PASSWORD}

kafka.properties — KRaft mode Kafka broker. Main values: process.roles=broker,controller, node.id=1, num.partitions=4, 3 types of listeners (SASL_PLAINTEXT/CONTROLLER/SASL_SSL), sasl.enabled.mechanisms=PLAIN, super.users=User:admin, allow.everyone.if.no.acl.found=false, log.dirs=${PP_DATA_DIR}/kafka/kraft_combined_logs, log.retention.ms=3600000 (1 hour), message.max.bytes=157286400, default.replication.factor=1, auto.create.topics.enable=true.

⚠️ log.retention.ms=3600000 = 1 hour. Based on the premise that Kafka acts as a buffer and persistence is handled by Aster TSDB/TSE. If a consumer stops for more than 1 hour, data is lost.

Plugin Template

plantpulse-plugin-opcua.properties — PostgreSQL / Aster TSDB / Valkey / Kafka connections + OPC UA server endpoint.

Property	Default	Description
kafka.topic	pp-tag-point	⚠️ Literal — not ${PP_TOPIC_PREFIX} substitution
kafka.thread · kafka.group.id	8 · opcua-plugin-group
opc.ua.server.domain · opc.ua.server.port	${PP_HOST_IP} · 8007
opc.ua.server.user · password · anonymous	opcua · 설치-시-변경 · false
opc.ua.server.tcp.port · tls.port	${PP_OPCUA_TCP_PORT:11004} · ${PP_OPCUA_TLS_PORT:11005}
redis.thread · redis.ssl.enabled	4 · false

plantpulse-plugin-aas.properties — AAS V3 server + Kafka consumption settings.

Property	Default	Description
aas.v3.port	${PP_AAS_V3_PORT:8090}	V3 HTTP port
kafka.topic	${PP_TOPIC_PREFIX}-tag-point	Real-time tag point reception
kafka.thread · kafka.poll.count	8 · 1000
group.id · auto.offset.reset · enable.auto.commit	pp-plugin-aas-server · earliest · false
session.timeout.ms · auto.commit.interval.ms	30000 · 1000
plugin.message.listener	plantpulse.plugin.opcua.messaging.KafkaPluginMessageListener

BaSyx V2 registry/AAS ports are hardcoded in code, not configured — REGISTRY_PORT=4800, AAS_PORT=4801 of AASServer.java.

---

Engine Keys Read by Code but Missing from Template

EngineConfig also reads the following keys. Since they are not in the template (plantpulse-engine.properties), code defaults apply normally. To change a value, you must directly add a line to plantpulse-server/config/plantpulse-engine.properties.

Edge Clock / Status Polling

Property	Code Default
edge.clock.poll.term.sec	60
edge.clock.skew.threshold.ms	10000
edge.clock.rtt.max.ms	2000
edge.clock.autosync.enabled	true
edge.clock.autosync.cooldown.ms	600000
edge.clock.warn.cooldown.ms	600000
edge.clock.skip.warn.count	5
edge.status.poll.seconds	30
edge.status.poll.jitter.max.ms	3000

edge.clock.poll.term.sec and edge.status.poll.seconds determine scheduler job cycles — refer to the Edge section of the scheduler architecture for value constraints.

DDS / Event Bus / Push / Buffer

Property	Code Default
engine.dds.queue.size	80000
engine.dds.worker.threads	4
engine.dds.latency.warn.ms	1000
engine.eventbus.queue.capacity	65536
engine.push.threads	4
engine.push.queue.size	10000
engine.storage.buffer.type	KAFKA
engine.streaming.timeout.backup.queue.capacity	12000000

The code default for engine.dds.enabled · engine.dds.data.tag.enabled · engine.dds.data.asset.enabled is false. Since the template specifies them as true, they are enabled only in the deployed version — deleting these three lines will silently disable DDS.

Job Thread (Not listed in template)

Property	Code Default
engine.job.thread.oee	10
engine.job.thread.ram	5
engine.job.thread.ems	5
engine.job.thread.system	4
engine.job.asset.slow.threshold.ms	20000

⚠️ The code default for engine.job.thread.point is 4, but the template is 8 (engine.job.thread.asset matches as 8 on both sides). Deleting the template line will halve the value.

Flow Engine (Not listed in template)

Property	Code Default
flow.engine.enabled	true
flow.executor.parallelism	64
flow.executor.queue.capacity	10000
flow.executor.queue.size	10000
flow.executor.poll.timeout.ms	50
flow.scheduler.pool.size	4
flow.timer.pool.size	2
flow.jdbc.trigger.pool.size	2
flow.shutdown.timeout.sec	5
flow.enabled.cache.ttl.ms	5000
flow.webhook.auth.enabled	true

Alarm / ISO Metric / OEE

Property	Code Default
alarm.duplicate.check.minutes	10
engine.oee.debounce.ms	100
iso.oee.rolling.window.minutes	10
iso.ems.rolling.window.minutes	10
iso.ram.rolling.window.minutes	1440

Anomaly Detection Details (Asset Point / System)

Meaningful only when enabled via engine.data.anomaly.enabled=true.

Property	Code Default
engine.data.anomaly.asset.point.interval	MINUTE
engine.data.anomaly.asset.point.lookback.minutes	14400
engine.data.anomaly.asset.point.recent.seconds	600
engine.data.anomaly.asset.point.limit	2100
engine.data.anomaly.asset.point.context.length	2048
engine.data.anomaly.asset.point.prediction.length	96
engine.data.anomaly.asset.point.score.history	60
engine.data.anomaly.asset.point.threshold.k	3.5
engine.data.anomaly.asset.point.threshold.min	0.80
engine.data.anomaly.asset.point.threshold.max	0.98
engine.data.anomaly.asset.point.topk.request	5
engine.data.anomaly.asset.point.topn.report	5
engine.data.anomaly.system.lookback.minutes	1440
engine.data.anomaly.system.recent.seconds	60
engine.data.anomaly.system.limit	2100
engine.data.anomaly.system.context.length	2048
engine.data.anomaly.system.pred.length	96
engine.data.anomaly.system.threshold.min	0.80
engine.data.anomaly.system.topk	2

Non-existent Settings

The following names appear in old documents or external materials but do not exist in templates or code. Adding settings with these names will have no effect.

Item sought	Actual location
storage.postgresql.*	metastore.driver / metastore.url / metastore.user / metastore.password
storage.redis.*	cache.host / cache.port / cache.password / cache.ssl.*
storage.cassandra.*	storage.host / storage.port / storage.keyspace / storage.user / storage.password
storage.tse.*	tse.protocol / tse.host / tse.port / tse.user / tse.password
storage.ttl.realtime etc.	storage.tag.point.ttl / storage.asset.data.ttl etc. per table (TTL section)
storage.compression.enable / .algorithm	storage.compression.zstd.level / storage.compression.zstd.type (Template lines are typos — Typo Key Warning)
storage.compaction.enable / .schedule	storage.table.compaction.strategy series (Compaction section)
storage.metastore.* / storage.cache.*	None
mq.type / mq.kafka.bootstrap.servers / mq.kafka.group.id	mq.host + mq.kafka.port. Consumer groups are code constants (PP-KAFKA-POINT-GROUP etc., Messaging Architecture), not configuration
mail.* in file	Only one place: pp.mm_config (Migrated 2026-09-17, file deleted)
websocket.* / plantpulse-websocket.properties	None. STOMP/WebSocket retired 2026-07, browser push is SSE (/push/sse) — Messaging Architecture
scheduler.enabled / scheduler.thread.count / scheduler.misfire.* / scheduler.history.*	None. Quartz configuration is plantpulse-server-web/src/quartz.properties (WAR classpath) — Scheduler Architecture
engine.ntp.enable / engine.ntp.server	engine.ntpdate.sync / engine.ntpdate.server.ip
engine.cep.enable / engine.cep.host / engine.cep.port	engine.cep.server.ip / engine.cep.server.port / engine.cep.server.protocol
engine.anomaly.*	engine.data.anomaly.*
engine.dds.enable / engine.dds.topic.prefix	engine.dds.enabled, Topic prefix is PP_TOPIC_PREFIX (mq template)
engine.async.queue.size / engine.async.timeout / engine.pipeline.batch.size / engine.timeout.* / engine.dataflow.threads / engine.dataflow.buffer.size	None
PP_STOMP_* / PP_WEBSOCKET_* env variables	Not in env.sh

Properties Containing Passwords

Values that must be replaced before delivery. Dev plaintext defaults are placeholders in env.sh (or template); for production/customer delivery, replace with customer-specific values or overwrite via Docker -e VAR=... / export.

File	Key
plantpulse-storage.properties	metastore.password · cache.password · cache.ssl.truststore.password · storage.password · tse.password · data.gateway.api.key
plantpulse-engine.properties	api.key · flow.webhook.api.key · engine.cep.server.api.key
plantpulse-mq.properties	mq.password · mq.ssl.truststore.password
pp.mm_config (Mail)	mail.smtp.password
pp.mm_config (AI)	ai.llm.api.key — No fallback for old name ai.openai.*
plantpulse-cep.properties	cache.password · storage.password · cep.api.key
plantpulse-batch.properties	metastore.password · cache.password · tse.password · mq.password · storage.db.password · analytics.password · external.db.password
plantpulse-jdbc.properties	cache.password · jdbc.postgres.password · jdbc.cassandra.password · jdbc.tse.password · jdbc.spark.password · data.gateway.api.key
plantpulse-data-gateway.properties	data.gateway.api.key
plantpulse-datalake-admin-api.properties	db_password · tse_password · redis_jmx_uri (password in URL) · redis_ssl_truststore_password · postgres_jmx_uri (password in URL)
plantpulse-mqtt.properties · plantpulse-mq-auth.properties	mqtt.password
plantpulse-plugin-opcua.properties	postgres.password · cassandra.password · redis.password · kafka.password · opc.ua.server.password
plantpulse-plugin-aas.properties	aas.db.password · cassandra.password · kafka.password
kafka.properties	ssl.keystore.password · ssl.key.password · ssl.truststore.password
hive-auth.properties	Entire value is password

API Key Policy: PP_API_KEY · PP_DATA_GATEWAY_API_KEY · PP_FLOW_WEBHOOK_API_KEY · PP_CEP_API_KEY use a dev plaintext default fallback (${VAR:-<dev-default>}) — this allows dev/test to boot and deploy without additional secret injection. Do not change to hard-fail (:?) — if restart-server.sh sources env.sh, it will crash immediately and deployment will fail.

Recommended Action

Secrets are not written in configuration files. Collect them in a single sidecar file and use 0600.

ls -l /etc/kopens/plantpulse-platform.env      # 비밀 — 0600, repo 밖
ls -l /etc/kopens/platform.node.env            # 이 박스 고유 값 — 0600

# 값을 바꿀 때는 파일을 열지 말고 도구를 씁니다
cd /opt/kopens/plantpulse-platform-docker
bin/passwd.sh --list                           # 바꿀 수 있는 키
bin/passwd.sh PP_PG_PASSWORD                   # 서버 계정 변경 → 사이드카 → 재렌더 → 재시작

Since the rendered configuration files contain the substituted values, do not commit those files to the repository.