Hi Luis, What we observed, is that with Pacific, tweaking > osd_mclock_max_capacity_iops_hdd, we can go from arround 200MB/s of writes > up to 600MB/s of writes, on balanced profile. > But with Reef, changing osd_mclock_max_capacity_iops_hdd does not change a > lot the performances of the cluster. (Or if it does, they are small enough > so I did not see them). > The above probably indicates that the default values for osd_mclock_max_capacity_iops_hdd are close enough to the actual capability of the backing device. > That been said, the performances of Reef "out of the box" are what we > expect of our cluster (arround 600MB/s), while with Pacific we needed to > tweak manually osd_mclock_max_capacity_iops_hdd to get the expected > performances. So there is definitely a big improvement there. > This is good feedback. One of our goals was to achieve a hands-free configuration of mClock and fine tune only when necessary. > > What made me think that this option was maybe not used anymore, during the > deploy of Pacific, each OSD pushes its own > osd_mclock_max_capacity_iops_hdd, but deploying Reef not. We did not see > any values for the OSDs in the ceph config db. > The fact that you don't see any values in the config db indicates that the default values are in effect. We added a fallback mechanism to use the default values in case the benchmark test during OSD boot-up returned unrealistic values. Please see https://docs.ceph.com/en/reef/rados/configuration/mclock-config-ref/#mitigation-of-unrealistic-osd-capacity-from-automated-test for more details and awareness around this. In your case, the configuration may be left as is since the defaults are giving you the expected performance. -Sridhar _______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
 |