Description of problem: There is a need for improvement through new features to make troubleshooting easier through the dynamic injection of parameters to modify the behavior of the software to provide the user with a quicker way to react in order to stabilize a cluster. The parameters that should be dynamically accessible include, but are not limited to: Pool migration scenario - To make troubleshooting much easier, being able to inject some timeout values. For example, the following settings are not injectable, which made it very difficult to get out of a cascading failure with a pool migration at a customer site: osd_op_thread_timeout=2000 filestore_op_thread_timeout=2000 osd_recovery_max_omap_entries_per_chunk = 1024 Cluster tuning/stabilizing - after or during a hardware failure or a pool migration: All scrubbing parameters All recovery and backfill parameters Having to set these commands and restarting thousands of OSD's is not practical and dynamic injection is required. Version-Release number of selected component (if applicable): 3.0
Updating the QA Contact to a Hemant. Hemant will be rerouting them to the appropriate QE Associate. Regards, Giri
*** Bug 1586204 has been marked as a duplicate of this bug. ***
Moving to 5.1, although it's not clear to me if we have concrete work here that we are committed to implement.