zeppelin flink interpreter in application mode cant find yarn resourcemanager

Question

I run on zeppelin (0.11.0) flink interpreter (flink 0.17.2). on yarn-session mode it works, but on yarn-application mode it stucks with:

INFO org.apache.zeppelin.flink.FlinkScala212Interpreter           [] - Use FlinkCluster in yarn application mode, appId: application_1731316069734_0048
INFO  org.apache.hadoop.yarn.client.RMProxy [] - Connecting to ResourceManager at /0.0.0.0:8032
INFO  org.apache.hadoop.ipc.Client [] - Retrying connect to server: 0.0.0.0/0.0.0.0:8032. Already tried 0 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)

As I see it, zeppelin flink RemoteInterpreterServer cannot find its configuration from yarn container, where its ran, and, thus, cannot find yarn RM. But, in first, flink job manager in this container succesfully initializes:

INFO org.apache.flink.configuration.GlobalConfiguration           [] - Loading configuration property: flink.yarn.resourcemanager.scheduler.address, ***.corpdomain:8030
...
INFO org.apache.hadoop.yarn.client.RMProxy                        [] - Connecting to ResourceManager at ***.corpdomain/10.9.12.200:8030

In second, when I run ./bin/flink run-application -t yarn-application ... (i.e. flink application mode without zeppelin interpreter payload) , it works with succeed.

How I must configure zeppelin flink interpreter for yarn-application mode? All settings from documentation I've done.

I tried to build and run zeppelin with hadoop support (USE_HADOOP=true), because in docs its required for 'interpreter yarn mode', but nothing is changed.