Ok,
After a general review of all the config files, I found the cause of the problem:
It was a misconfiguration in the cluster name declared in hdfs-site.xml in the mentioned entry that should write
<name>dfs.client.failover.proxy.provider.qbs-cluster</name>
instead of
<name>dfs.client.failover.proxy.provider.qbs-mycluster</name>
After correcting that mistake, I could create all mapred related folders and start the jobhistory successfully.
Hope this could help others with the same problem to debug it faster.
Regards,
SysAdmin