Changed IP on Cluster (SOLVED)

[JackieBrown]

I am getting a new router and figured it would be a good time to orgainze all my IP addresses.

I changed the IP addresses on the cluster updating
/etc/network/interfaces
/etc/hosts
/etc/pve/corosync.conf

This seemed to have broken my cluster.
Node 3 can see node 4
Node 2 can see node 5
Node one only works if I switch to local only.

Here is my coronsyc file

logging {
  debug: off
  to_syslog: yes
}

nodelist {
  node {
    name: hp1
    nodeid: 4
    quorum_votes: 1
    ring0_addr: 192.168.1.10
  }
  node {
    name: hp2
    nodeid: 3
    quorum_votes: 1
    ring0_addr: 192.168.1.11
  }
  node {
    name: pve
    nodeid: 1
    quorum_votes: 1
    ring0_addr: 192.168.1.14
  }
  node {
    name: razendesk
    nodeid: 2
    quorum_votes: 1
    ring0_addr: 192.168.1.13
  }
  node {
    name: router
    nodeid: 5
    quorum_votes: 1
    ring0_addr: 192.168.1.12
  }
}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: MyNewServer
  config_version: 15
  interface {
    linknumber: 0
  }
  ip_version: ipv4-6
  link_mode: passive
  secauth: on
  version: 2
}

 

[JackieBrown]

Edit - All nodes but node 1 can see each other now but they do seem to randomly fall off. I do not see any configuration different from node 1 versus the others, though

 

[jsterr]

Did you just try to replace the ips in corosync.conf without removing the interfaces before? You will now need: https://forum.proxmox.com/threads/made-mistake-in-corosync-conf-now-cannot-edit.77173/#post-716686 you are having a split-brain-scenario now. It is not allowed to edit active links and give them a new ip-adress. you need to add a new interface and the remove the old. Changing an already active corosync-interface will result in split-brain.

Your journalctl -u corosync.service should show those errors.

I would also recommend adding a second link generally, in case ring0 fails, you will get reboots because quorum is lost.

 

[JackieBrown]

By new interface, you mean under totem? Change the link number?
also, when I run
corosync-cfgtool -s
on the node that does not seem to be connected, it shows everything connected via the cml.

root@pve:~# corosync-cfgtool -s
Local node ID 1, transport knet
LINK ID 0 udp
addr = 192.168.1.14
status:
nodeid: 1: localhost
nodeid: 2: connected
nodeid: 3: connected
nodeid: 4: disconnected
nodeid: 5: connected

 

[jsterr]

By new interface, you mean under totem? Change the link number?

No i mean the ring-IP-Adresses. its not allowed to change the ips in here. you need to first add a ring1, reload config. then remove ring0. You cant just simply switch the ip-adresses here.

 

[JackieBrown]

That seem to have fixed it. I'll run it for a bit and see if any fall off again.

Thank you