rook: manually repair OSD after rook cluster fails after k8s node restart

I was faced with the problem of falling OSD after restarting any node k8s. It is alleged that the OSD needs to stand up for himself, But it’s not working for me. The problem is described in more detail here: https://github.com/rook/rook/issues/1278

And here I want to discuss possible ways to get OSD to work manually.

ceph osd tree
ID  CLASS WEIGHT  TYPE NAME           STATUS REWEIGHT PRI-AFF
 -1       0.46196 root default
 -2       0.09239     host 10-1-29-31
  1   hdd 0.09239         osd.1           up  1.00000 1.00000
-11       0.09239     host 10-1-29-32
  4   hdd 0.09239         osd.4           up  1.00000 1.00000
 -3       0.09239     host 10-1-29-33
  0   hdd 0.09239         osd.0           up  1.00000 1.00000
 -9       0.09239     host 10-1-29-34
  2   hdd 0.09239         osd.2         down        0 1.00000
 -4       0.09239     host 10-1-29-35
  3   hdd 0.09239         osd.3           up  1.00000 1.00000

I have tried the following methods found in the documentation:

ceph osd repair osd.2
Error EAGAIN: osd.2 is not up

ceph osd up osd.2
no valid command found; 10 closest matches:
osd count-metadata <property>
osd versions
osd find <osdname (id|osd.id)>
osd metadata {<osdname (id|osd.id)>}
osd getmaxosd
osd ls-tree {<int[0-]>} {<name>}
osd getmap {<int[0-]>}
osd getcrushmap {<int[0-]>}
osd tree {<int[0-]>} {up|down|in|out|destroyed [up|down|in|out|destroyed...]}
osd ls {<int[0-]>}
Error EINVAL: invalid command

ceph-volume simple activate osd.2
\-->  RuntimeError: Expected JSON config path not found: /etc/ceph/osd/osd.2-None.json

all of them lead to errors and do not give the expected result. maybe I’m moving in the wrong direction?