> On 19 Jul 2019, at 15:25, Sage Weil <[email protected]> wrote: > > On Fri, 19 Jul 2019, Stig Telfer wrote: >>> On 19 Jul 2019, at 10:01, Konstantin Shalygin <[email protected]> wrote: >>>> Using Ceph-Ansible stable-4.0 I did a rolling update from latest Mimic to >>>> Nautilus 14.2.2 on a cluster yesterday, and the update ran to completion >>>> successfully. >>>> >>>> However, in ceph status I see a warning of the form "Legacy BlueStore >>>> stats reporting detected” for all OSDs in the cluster. >>>> >>>> Can anyone help me with what has gone wrong, and what should be done to >>>> fix it? >>> I thin you should start to run repair for your OSD's - [1] >>> [1] >>> http://lists.ceph.com/pipermail/ceph-users-ceph.com/2019-July/035889.html >>> <http://lists.ceph.com/pipermail/ceph-users-ceph.com/2019-July/035889.html> >>> k >>> >> Thanks Konstantin - running the BlueStore repair as indicated appears to >> be working. >> >> One difference from Sage’s description of the scenario is that I did not >> explicitly create new OSDs during or after the Nautilus upgrade. Perhaps >> the Ceph-Ansible rolling update script did something to trigger this. > > The warning is just telling you there are legacy bluestore instances that > were created before nautilus. It is perhaps a bit harsh--you can also > just silence the warning if you don't care about getting the newer > and more accurate per-pool stats. > > sage
Thanks Sage and Paul for clarifying that - good to know. Best wishes, Stig _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
