@Yasuhiro Shindo Container killed exit code most of the time is due to memory overhead If you haven't specified spark.yarn.driver.memoryOverhead or spark.yarn.executor.memoryOverhead these params in your spark submit then add these params (or) if you have specified then increase the already configured value. Please refer to this link to decide overhead value. ... View more

@Alvarez Rafa QueryDatabaseTable processor stores the state when processor ran for the first time based on Max value column(idmovil). For the next run processor only pulls the changes from the table based on idMovil column. You can check the last state value by RightClick on processor -> View State To clear the state stop the processor and RightClick on processor -> View State and then clear state. Once we clear the state then processor pulls all records from the table. --- If you are not facing this issue then please attach your flow screenshot and scheduling on QueryDatabaseTable processor. ... View more

@HanYan Tan Could you look into zeppelin-4140 jira is reported for the same issue. Checkout the comments associated with the jira for more details, as mentioned in jira comments "jdbc interpreter Binding Mode Per User Scoped and isolated mode per user the temporary table is dropped" Try with the above setting and check does the issue resolved or not 🙂 ... View more

@Raj Negi After FetchHbaseRow processor use ReplaceText processor with below configs: Search Value (?s)(^.*$) Replacement Value ${'$1':unescapeJson()} //capture all the data and apply nifi expression language unescapeJson function. Character Set UTF-8 Maximum Buffer Size 1 MB //change as per your flowfile size Replacement Strategy Regex Replace Evaluation Mode Entire text Flow: 1.FetchHbaseRow 2.ReplaceText --other processors ... View more

@Kevin Lahey Not sure if you are using NiFi cluster (or) not, could you try to run ListS3 processor only on Primary Node only. As this processor intended to run only on primary node as per documentation. ... View more

@john y We can directly access filesize with ${fileSize} and this attribute expression will result the actual filesize value of flowfile. ... View more

@Manish Parab Your GetMongo processor is running on All nodes that means same data is pulled on all nodes.. If you select EvaluateJsonPath processor to run on Only primary node then all the other nodes flowfiles will be left queue before EvaluateJsonPath processor, Because you are not processing flowfiles that are pulled from all other nodes except of PrimaryNode. Run GetMongoProcessor only on primary node and keep EvaluateJsonPath Processor to run on all nodes, Reason to keep EvaluateJsonPath processor on all nodes if NiFi primary node changed then EvaluateJsonPath processor not going to processor the flowfile that are listed on old PrimaryNode. ... View more