Please provide the internal manager log file so we can check.
Also you can try below action plan.
1) check below parameter "_lm_global_posts" , "_immediate_commit_propagation" valuse set to 'true' in database.
2) do node fnd node cleanup
3) run autoconfig in all nodes.
4) do cmclean.sql
5) start the application and check.
"adcmctl.sh stop on node 2, all the managers in Node 1 are stopping". --> Issuing the command "adcmctl.sh stop" on one server will stop the managers on all the servers configured with CP services. This is the expected behavior.
Note: If any of the replies in this thread were helpful/correct, kindly mark it accordingly as this would help others who can also benefit from it.