Luckily some riak_tests failed when upgrading from riak_ee to riak OS next. JMX config was why.
Basho's old Riak EE came with replication, JMX, and SNMP. bet365 opensourced all these, and have added replication to open source Riak. Running the replication riak_test suite leads to node start up failures when upgrading in place from Riak EE to OS Riak >= 2.2.4. The failures are caused by existing JMX properties from cuttlefish's inclusion of the JMX schema file when it generated the Riak EE riak.conf. I have no idea if this would be an issue in a real life upgrade scenario (do people keep their riak.conf between upgrades?) Here are the options as I see it, if you think of more and better, say so, please.
• Do not add JMX and SNMP to OS Riak, and document that you need to remove all JMX/SNMP properties from riak.conf files before starting up. People will notice if their nodes don't start.
• Do not add JMX and SNMP to OS Riak and add a bunch of dummy JMX/SNMP properties to riak_kv's schema file. On start up, detect these properties and log an ERROR that JMX/SNMP are not supported.
• Add JMX and SNMP to OS Riak and both document and log a warning that they're deprecated. Remove after 2.2.5