- Publishing 5000+ pages can be done successfully, but when publishing 18,000+ pages deadlock errors begin to occur and the Deployers shut down. - Deployer errors seen are ERROR SqlExceptionHelper - Transaction (Process ID 77) was deadlocked on lock resources with another process and has been chosen as the deadlock victim. Rerun the transaction. ERROR ExceptionMapperStandardImpl - HHH000346: Error during managed flush [org.hibernate.exception.LockAcquisitionException: could not execute statement] WARN StepActor - Step 'TridionUnzipStep' for 'tcm:0-5218799-66560' failed with message: org.springframework.dao.CannotAcquireLockException: could not execute statement; SQL [n/a]; nested exception is org.hibernate.exception.LockAcquisitionException: could not execute statement ERROR StepActor - Step 'TridionUnzipStep' for 'tcm:0-5218799-66560' can NOT retry ERROR StepActor - Step 'TridionUnzipStep' for 'tcm:0-5218799-66560': org.springframework.dao.CannotAcquireLockException: could not execute statement; ... WARN TridionCleanupStep - Could not locate original zip location for execution id tcm:0-5218799-66560 WARN StepActor - Step 'ComponentDeploy' for 'tcm:0-5224239-66560' failed with message: org.springframework.beans.factory.BeanCreationNotAllowedException: Error creating bean with name 'transactionManager': Singleton bean creation not allowed while the singletons of this factory are in destruction (Do not request a bean from a BeanFactory in a destroy method implementation!) ERROR TransactionInterceptor - Application exception overridden by rollback exception org.hibernate.LazyInitializationException: failed to lazily initialize a collection of role: com.sdl.delivery.deployer.jpa.domain.StepEntity.executionProperties, could not initialize proxy - the owning Session was closed ... WARN StepActor - Step 'TridionCleanupStep' for 'tcm:0-5224282-66560' failed with message: java.lang.IllegalStateException: Session/EntityManager is closed- Deployer JVM configuration was JVM_OPTIONS="-Xrs -Xms512m -Xmx1024m -Dfile.encoding=UTF-8"- Deployer trace files are regularly removed as discussed in KB article SDL Web - How to cleanup files left over by publishing - Below SQL was run on broker DB to change isolateion mode (or on state store DB if separate) DELETE FROM [Broker_Stg].[dbo].[EXECUTION] ALTER DATABASE Broker_Stg SET ALLOW_SNAPSHOT_ISOLATION ON ALTER DATABASE Broker_Stg SET READ_COMMITTED_SNAPSHOT ON- Hotfix CD_8.5.0.11391 had been applied. |
Issue resolved after increasing the default Xmx setting from 1024 to 2048, with no further crashes of microservices seen when mass publishing. |