ceph gui gets timeout

[hesmebye]

Hi,

I am very new to proxmox and even newer to ceph.
I had three nodes in my cluster, and ran the ceph installation from the GUI. This installed the software on my nodes, and probably startes some services (I did not go any further than installl the software) This was just for getting familiar with the GUI, and I did not created any disk/storage at this point.

Then I added two new nodes and removed two of the three in the original cluster. So my cluster now consist of one node that has been in the cluster all along, and two new nodes where I have not installed ceph.
After adding and removing the new nodes I have added one disk on each node and created a separate 1gb nework - so no I am ready for actually installing ceph and do some testing.

The problem is - on all the nodes the ceph proxmox gui gets a timeout, and there is not possible to install ceph on the two latest nodes.
So I basicaly want to remove ceph on all nodes, and start at scratch. I am running proxmox 7.3.3 and I installed the latest version of ceph.

The web gui in proxmox seem to do a GET to https://192.168.1.177:8006/api2/json/nodes/pve3/ceph/status which results in a 500.

How do I get rid of all traces of ceph, and how do I reset my gui so that I can install everything again --
or should I install ceph from the commandline and continue from there?

I have read some thread about stopping and removing the osd, mons and managers services etc, but I cant find any such services on my machines, not even the one that was part of the cluster all along. So what is actually running / getting the gui to behave like this and how to fix it?

Thanks
Helge Smebye

 

[rason]

To remove all traces of Ceph from your Proxmox nodes, you can follow these steps:

1. Stop all OSDs on all nodes using the Proxmox GUI or the command line.
2. Remove all OSDs using the Proxmox GUI or the command line.
3. Remove all Mons (except the master) using the Proxmox GUI or the command line.
4. Remove all Managers (except the master) using the Proxmox GUI or the command line.
5. Execute pveceph purge on each OSD node using the command line.
6. On the last node (master mon/mgr), stop all Ceph services and execute pveceph purge using the command line.
7. Finally, remove all Ceph configuration files located at /etc/ceph/ and /var/lib/ceph on all nodes using the rm -rf command.

After completing these steps, you should have removed all traces of Ceph from your Proxmox nodes and can now install Ceph again from either the GUI or the command line. Remember to remove any remaining configuration files before starting the installation process to ensure a clean setup. Good luck!

 

[hesmebye]

Thanks for a quick response!

None of the deamons where running, but I still had problems with the gui in proxmox.
The node that was in the cluster from the beginning had a ceph.conf under /etc/pve. The existance of this config file triggered the gui to seach for https://192.168.1.177:8006/api2/json/nodes/pve3/ceph/status

By removing the ceph.conf file under /etc/pve/ on this machine, the gui stopped searching for the cluster.
So
mv /etc/pve/ceph.conf /etc/pve/ceph.conf-old
was the solution on the problem.