Broker running on 8 vcpus VM with the below config. Observing high CPU usage and the broker service restarts, no JVM crash logs. What could be the cause ?
Server Configuration
druid.server.http.maxIdleTime: PT5M
druid.server.http.enableRequestLimit: false
druid.server.http.defaultQueryTimeout: 300000
druid.server.http.gracefulShutdownTimeout: PT0S
druid.server.http.unannouncePropagationDelay: PT0S
druid.server.http.maxRequestHeaderSize: 8192
druid.server.http.numThreads: 40
Client Configuration
druid.broker.http.numConnections: 20
druid.broker.http.compressionCodec: gzip
druid.broker.http.readTimeout: PT15M
druid.broker.http.unusedConnectionTimeout: PT4M
druid.broker.http.maxQueuedBytes: 10485760
Retry Policy
druid.broker.retryPolicy.numTries: 1
Processing
druid.processing.buffer.sizeBytes: 536870912
druid.processing.buffer.poolCacheMaxCount: 2147483647
druid.processing.formatString: processing-%s
druid.processing.columnCache.sizeBytes: 0
druid.processing.fifo: false
druid.processing.numMergeBuffers: 3
Segment Discovery
druid.serverview.type: batch
druid.broker.segment.watchedTiers: none
druid.broker.segment.watchedDataSources: none
druid.broker.segment.awaitInitializationOnStart: true
Query cache
druid.broker.cache.useCache: true
druid.broker.cache.populateCache: true
druid.cache.type: local
druid.cache.sizeInBytes: 2000000000
JVM Runtime Properties
jvm.config.Xms: Xms4g
jvm.config.Xmx: Xmx8g
jvm.config.XXOptions: XX:MaxDirectMemorySize=8g -XX:+ExitOnOutOfMemoryError -XX:ErrorFile=/logs/druid/hs_err_pid%p.log