Multiple warnings can occur in the normal situation of taking a node down. For example

LOGGER = com.arjuna.ats.jta.logging.loggerI18N
NATIVE_LEVEL_NAME = WARN
MESSAGE = [com.arjuna.ats.internal.jta.recovery.xarecovery1] Local XARecoveryModule.xaRecovery got XA exception javax.transaction.xa.XAException: Error trying to connect to any providers for xa recovery, XAException.XAER_RMERR

LOGGER = org.hornetq.core.client.impl.ClientSessionFactoryImpl
NATIVE_LEVEL_NAME = WARN
MESSAGE = Tried 1 times to connect. Now giving up on reconnecting it.

On node is complaining about the other node not being there. Probably an info would be enough in this case.

This appears to be logged as a warning because it might not be normal not to be able to talk to the other server. But if the component (for instance arjuna recoverer) was checking with the cluster component, it would see that the other node is not up, so it could just assume that this is a normal situation.
It should log a warn if the component cannot do the recovery that it should be able to (for instance when the cluster component says that the node is up).