We had a cluster with version 5.5 of kerberos (8 nodes). On the cluster, tasks in oozie were successfully running. We updated the cluster to version 5.13. After that, an error began to appear. And the task can then run back, then after, for example, 8 times not to run back. Error: Main class [org.apache.oozie.action.hadoop.Hive2Main] exit code [2].
There are no details in the oozie and sheduler logs.
When testing errors it was revealed:
- Occurs when the hive2action block is executed.
- At numerous attempts, it was revealed that out of 10 attempts approximately 1 time the script runs off. In this case, the input data, connection logs of libraries, etc. were compared and were identical.
- Oozie script is written correctly, all dependencies are used.
During testing, 3 run statuses were identified (two that fell with errors and one ran away): no connection1-2.jpg, interrupted connection1-2.jpg, success connection1-2.jpg.
1. success connection - the script ran away. In the second screenshot, you can see that the connection went to beeline and !сonnect... with the appropriate parameters, and then the insert ran off, which is shown in the logs on screenshot 1.
2. no connection - the script fell. On the 2nd screenshot you can see that there is an attempt to connect to the beeline, as if !сonnect... parameters were not applied. BUT, these parameters (url, principal) were passed and this can be seen in the oozie coordinator of the web-face of this oozie action.
3. interrupted connection - the script fell. In the 2nd screenshot, you can see that the parameters were connected to the beeline with the appropriate! Connect parameters, but then it seemed as if there was a break. And from the 1st screenshot you can see that the error “Failing Oozie Launcher, Main Class [org.apache.oozie.action.hadoop.Hive2Main], exit code [2]” also went
Please help!
