I have several Sqoop jobs running to pull data from Oracle data source. Ideally, these jobs are running fine but sometimes I have seen that few tables are getting stuck in the accepted stage even after sufficient resources are being allocated. The resource allocations are as follow:
App type: MAPREDUCE, elapsedTime: 11h 12m 51s, Queue: ingestion, Queue usage percentage: 1.025 %, Allocated VCores: 2, Allocated Memory: 12.29 GB
This table dont have much record count and generally, this completes in 10 min but today it took 11h with no progress and even no data is being fetched from the source. I am trying to debug and fix this issue and looking for assistance on this.
It seems like you would need to look at the yarn queue whether this job was launched and see from the RM UI to understand if the queue was over utilized or not. That's the only way to identify if there was a resource contention and due to which the app stayed in accepted state.