For testing purposes, we are working on the assumption that this bug is caused by cache invalidation notifies failing to propagate to all machines under heavy load Thus, as with the other bug (as discussed before with me and Eric) run the cluster under heavy enough load with multiple gateways until the log starts showing failures to propagate cache. Have the gateways update bucket metadata. Eventually you should be able to provoke the behavior where one node has failed to have its cache updated and responds with a 500 error. The fix should make that not happen
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:0474