TimeoutException with taskLib

Enonic version: 7.7.3

I’m having the following error when calling taskLib methods in a QA server

    MemberCallableTaskOperation failed to complete within 5000000000 NANOSECONDS.
     Invocation{op=com.hazelcast.executor.impl.operations.MemberCallableTaskOperation{serviceName='hz:impl:executorService', identityHash=1051831816, partitionId=-1, replicaIndex=0, callId=6130, invocationTime=1637257264218 (2021-11-18 17:41:04.218), waitTimeout=-1, callTimeout=60000, name=xp/task}, tryCount=250, tryPauseMillis=500, invokeCount=1, callTimeoutMillis=60000, firstInvocationTimeMs=1637257264385, firstInvocationTime='2021-11-18 17:41:04.385', lastHeartbeatMillis=0, lastHeartbeatTime='1970-01-01 00:00:00.000', target=[]:5701, pendingResponse={VOID}, backupsAcksExpected=0, backupsAcksReceived=0, connection=null}

This error occurs only in the QA server, in another server with the same XP version the methods work fine. This also happens only in the initialization of the application. When a task is called after that, in a service, for example, it works fine too.
We are using the same version for Enonic and taskLib.
Any ideas of what could be causing this?

Probably your QA cluster node is running out of memory and spend all time running Java GC?
In this case task is sent to a node, but not accepted on time