1

I have a high availability solution which requires me to have all my nodes using the same hostname. I've been tasked to use ceph cluster storage, but I just reached a dead end.

I can't figure out how to configure ceph so that I can use the same hostname for different nodes, because I found out that ceph nodes names must match the hostnames of the machines they belong to, source. Can anyone give me advice or show me how I can do so ?

5
  • What on earth could possibly require you to have multiple nodes with the same hostname?! Jul 14, 2020 at 16:03
  • @MichaelHampton the machines are connected to another machine that manages them using their hostname, the managing machine reaches the master node using a virtual IP address, if the master dies then the backup machine becomes master and can start responding to the managing machine. The clients also don't know that there are multiple nodes, they are connected using the same virtual IP. Also there is a health check done using the managing machine that needs a hostname ( the whole topology wouldn't be possible if there are more that 1 hostnames)
    – Malazzar
    Jul 14, 2020 at 16:13
  • 1
    That doesn't explain why each node has to have the same hostname. The common hostname can simply refer to the virtual IP, which is how it's normally done. Jul 14, 2020 at 16:16
  • @MichaelHampton the managing machine would need to have a second entry for the other node. The point is that the managing machine also doesn't know about it. Lets say I want to update the nodes using this manager, if I do so using 1 entry on the master node then it's just 1 query. The slave node will then replicate from the master and all is well. Otherwise I would need to do the same thing for the other node(another entry) . The manager's entry for these nodes is dependent on the hostname.
    – Malazzar
    Jul 14, 2020 at 16:22
  • Eh? It sounds like you are using some sort of homebrew HA script rather than a well tested solution like pacemaker or keepalived? Jul 14, 2020 at 16:26

1 Answer 1

0

I have a high availability solution which requires me to have all my nodes using the same hostname

Ceph will not work like this, ever, nor would anything else acting as a cluster, or anything else vaguely sensible either for that matter. Just do what 100% of the rest of the world does and build workable systems.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .