about ceph RBD

[lincarry]

My cluster:

Carry-1 carry-2 carry-3

Each node has an OSD

3 OSD generates RBD (size3 min_size1 pg_num 64)

When I simulate a fault to close two nodes, the RBD has a question mark and ceph -s does not respond. Why is this happening? If there is only one cluster left, what should I do to make sure the cluster is working properly?

 

[lincarry]

a student sincerely listens to your teaching

 

[sg90]

Do you have a mon installed on each server?

It's sigested never to run CEPH with a min of 1.

 

[lincarry]

Do you have a mon installed on each server?

It's sigested never to run CEPH with a min of 1.

Mon, mds, mgr have been created for all nodes

 

[sg90]

Mon, mds, mgr have been created for all nodes

Then your have 2 out of 3 monitors down theirfore you can't get a quorum.

 

[lincarry]

Do you have a mon installed on each server?

It's sigested never to run CEPH with a min of 1.

my proxmox version is 6.0.2.

Is it related to this parameter?

 

[sg90]

No, nothing to do with Proxmox.

Ceph Mon's run in a quorum , with a quorum then CEPH will not operate at all.

With only 1 out of 3 mon's are online CEPH will become unavailable as your seeing.

 

[lincarry]

No, nothing to do with Proxmox.

Ceph Mon's run in a quorum , with a quorum then CEPH will not operate at all.

With only 1 out of 3 mon's are online CEPH will become unavailable as your seeing.

I think I understand, thank you very much.