Travis Nielsen

The [default for the dashboard is `ssl: true`](https://github.com/rook/rook/blob/master/deploy/charts/rook-ceph-cluster/values.yaml#L128). Is it possible you didn't override that default during the upgrade? If it reverted to the default, then I could see this...

and please ignore my comment in our separate discussion that this overlapped with #10956, it's really a separate feature, thanks.

Ceph will create the `.nfs` pool automatically if there is no pool CR defined for it, so you're correct no need to create it explicity. @jmolmo Ideas on this dashboard...

What is the status of the `PersistentVolume`s created in cluster-on-local-pvc.yaml? They must not be binding to the PVCs requested by Rook for the OSDs. Did you also update the PVs...

To move ahead with the service account solution, instead of waiting on the operator sdk dependency mentioned in #6673, I'd suggest we bring back #7406 and add a workaround as...

> > To move ahead with the service account solution, instead of waiting on the operator sdk dependency mentioned in #6673, I'd suggest we bring back #7406 and add a...

> A few of us worked around it by using the default in the helm chart instead of setting it to 17.2. Seems to be a bug in that version...

If you're running rook v1.9.3, this may be the same cause as #10219.

> I think this commit (#11109) will resolve the osdprepare memory limit issue. #11109 was included in v1.10.3 and v1.9.12. Removing the memory limits on the osd prepare is expected...

The log shows the devices can't create new OSDs: ``` 2022-09-08 01:33:24.709844 I | cephosd: skipping device "sdc", detected an existing OSD. UUID=56938a4b-4d23-4679-863b-4a6762bc5ffd 2022-09-08 01:33:24.709972 I | cephosd: skipping device...