Description of problem (please be detailed as possible and provide log snippests): After upgrading cluster to 4.14, deviceClasses is still set to hdd instead of ssd. Also "ceph osd tree" is dtill showing as hdd Version of all relevant components (if applicable): ODF upgraded from 4.13.0-218 to 4.14.0-90 Does this issue impact your ability to continue to work with the product (please explain in detail what is the user impact)? No Is there any workaround available to the best of your knowledge? No Rate from 1 - 5 the complexity of the scenario you performed that caused this bug (1 - very simple, 5 - very complex)? 1 Can this issue reproducible? 1/1 Can this issue reproduce from the UI? Not Tried If this is a regression, please provide more details to justify this: Steps to Reproduce: 1. install ODF 4.13.0-218 using ocs-ci 2. upgrade cluster to 4.14.0-90 3. check ssd is updated in all places ( cephcluster, "ceph osd tree" ) Actual results: $ oc get cephcluster ocs-storagecluster-cephcluster -o yaml | grep -i deviceClasses -A1 deviceClasses: - name: hdd $ $ oc exec rook-ceph-tools-796c9b77d6-ms64z -- ceph osd tree ID CLASS WEIGHT TYPE NAME STATUS REWEIGHT PRI-AFF -1 1.50000 root default -12 0.50000 rack rack0 -11 0.50000 host ocs-deviceset-0-data-0fksmt 2 hdd 0.50000 osd.2 up 1.00000 1.00000 -4 0.50000 rack rack1 -3 0.50000 host ocs-deviceset-1-data-0qxmqx 1 hdd 0.50000 osd.1 up 1.00000 1.00000 -8 0.50000 rack rack2 -7 0.50000 host ocs-deviceset-2-data-0sp4kp 0 hdd 0.50000 osd.0 up 1.00000 1.00000 Expected results: it should show ssd in all places Additional info: upgrade job: https://url.corp.redhat.com/e054a89 must gather: https://url.corp.redhat.com/dc9ef4f
Hi Vijay This behavior is expected. The change was only for new clusters/OSDs. More info about this can found in the EPIC description and comments - https://issues.redhat.com/browse/RHSTOR-4326
Moved to on_QA in case the QE as any concerns. Please close the BZ if there are no further concerns.
(In reply to Santosh Pillai from comment #2) > Hi Vijay > > This behavior is expected. The change was only for new clusters/OSDs. More > info about this can found in the EPIC description and comments - > https://issues.redhat.com/browse/RHSTOR-4326 There is inconsistency here. Its updated for the "crushDeviceClass" and "tuneFastDeviceClass" parameters where as in other places ( mentioned in description ) its not updated. We need to decide whether we update at all places or not update anything. cephcluster: https://url.corp.redhat.com/7ebca76
(In reply to Vijay Avuthu from comment #4) > (In reply to Santosh Pillai from comment #2) > > Hi Vijay > > > > This behavior is expected. The change was only for new clusters/OSDs. More > > info about this can found in the EPIC description and comments - > > https://issues.redhat.com/browse/RHSTOR-4326 > > There is inconsistency here. Its updated for the "crushDeviceClass" and > "tuneFastDeviceClass" parameters where as in other places ( mentioned in > description ) its not updated. > We need to decide whether we update at all places or not update anything. > > cephcluster: https://url.corp.redhat.com/7ebca76 Hi Chris, Any suggestions on this? We expected this because rook does not change the deviceClass of existing OSDs. Thanks, Santosh
yes, please keep the device class of existing OSDs on upgrade the way it was before the upgrade. Only new OSDs that are added should be marked SSD. This means the following can be possible: 1) ODF 4.13, all OSDs marked as HDD 2) Upgrade to ODF 4.14, all OSDs still marked as HDD 3) Scale out, new OSDs are marked as SSD, old as HDD