The multinode sessions stop working when having to switch to a remote node.
The uprocs having a dependency condition on a remote Event, remain in status Event Wait.
Error Message :
These kind of errors appear on the universe.log
######################################################
ERROR|X|IO |pid=7692.7404| owls_api_exchanger_create | Error 200 connecting to node <NODENAME> port 10600
ERROR|X|IO |pid=2564.3000| k_handshakeAuthent | authentication to []/[NODE_HOSTNAME] failed, status [3]
ERROR|X|IO |pid=2564.3000| owls_api_exchanger_create | Error 300 connecting to node <NODENAME> port 10600
ERROR|X|IO |pid=2564.3000| o_module_exc_cycle_emissi | Cannot send exchanger message to remote node NODENAME
######################################################
These errors may also appear:
On Windows node:
######################################################
|ERROR|X|IO |pid=6176.8872| fopen | Error opening file [D:\duas_folder\data\exp\network\0000001.receive] mode (rb): No such file or directory
######################################################
On Linux remote node:
######################################################
ERROR|X|IO |pid=95966.139829996988736| fopen | Error opening file [/duas_folder/data/exp/fla/0036727.input] mode (rb): No such file or directory
######################################################
OS: All
Cause type: Configuration
Root Cause: The remote node had several IP addresses and the one used to declare the node on the UVMS (NODE_HOSTNAME) pointed to an IP address which was not allowed on the firewall for the local node to communicate with.
In order to fix the issue, add a line on the hosts file of the local node pointing to the IP address of the remote node that is opened on the firewall.
A better solution would be asking the Network team to create a different alias on the DNS pointing to that IP address and the modify the declaration of the remote node via the command:
unims -update -host NEWHOSTNAME
A restart of the node is necessary to take into account the change.