problem Cluster and CEPH

[Daly]

hi,

i had cluster with 3 nodes , i delnode the third node and remove crush osd after that i reinstall proxmox but when i add it to the cluster i can't create the 4 osd that i already delete.

 

[sb-jw]

Have you ever wiped disks?

 

[Daly]

yes i wiped disks

 

[sb-jw]

OK. But how exactly do you know that you can no longer create the discs as OSD?

 

[Daly]

i can't select my disks when i'm tryin to create osd.
i must remove VMs disk from storage?

 

[sb-jw]

To do this, you have to be root and have selected the corresponding node on the left, then go to the CEPH and there to the OSDs.

Otherwise, please click on Disks for the node and post a screenshot.

 

[Daly]

here is screenshot from the left node

 

[sb-jw]

So first of all, this is not the node, you have selected a storage. The node is the one above and then where VM Disks is now there is also the CEPH selection and right there in the submenu there is also OSD.

Second, your cluster doesn't seem to be a cluster. Node 3 has a green tick, the other two have none. It looks like you haven't reinstalled the node and old data may be present.
Or it just looks like that in the picture and the hooks are reloaded.

However, as long as it is not clear to us that your cluster is really running smoothly, we don't need to worry about the CEPH issue for the time being.

 

[Daly]

I have reinstalled node 3, so currently, node 3 has not yet joined the cluster. so firstly i must join the cluster then install ceph then create osd to sync with other osd?
attached screenshot of the node 3

 

[sb-jw]

Basically, that should be the way, yes. However, it may be that fragments still exist somewhere and you get error messages here and there.

But you just have to try it now, only then will we know.

 

[Daly]

here what i get when i'm tryin to create osd "Connection error 599: Too many redirections"
also when i'm ryin to create montor ceph : "Could not connect to ceph cluster despite configured monitors (500)"