(UDPU) Waiting for Quorum (Stucked)

[masterdaweb]

I have a cluster with 3 nodes working smoothly, I'm using UDPU instead of Multicast.

When I try to add the 4th node, it hangs on Waiting for Quorum.

Here's is my Corosync:

logging {
  debug: off
  to_syslog: yes
}

nodelist {
  node {
    name: ns524364
    nodeid: 1
    quorum_votes: 1
    ring0_addr: ns524364
  }

  node {
    name: ns545353
    nodeid: 4
    quorum_votes: 1
    ring0_addr: ns545353
  }

  node {
    name: ns541194
    nodeid: 3
    quorum_votes: 1
    ring0_addr: ns541194
  }

  node {
    name: ns535698
    nodeid: 2
    quorum_votes: 1
    ring0_addr: ns535698
  }

}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: mamae
  config_version: 62
  ip_version: ipv4
  secauth: on
  transport: udpu
  version: 2
  interface {
    ringnumber: 0
  }

}

 

[Alwin]

The udpu transport needs a list of members. See this example.
https://linux.die.net/man/5/corosync.conf

 

[t.lamprecht]

I have a cluster with 3 nodes working smoothly, I'm using UDPU instead of Multicast.

When I try to add the 4th node, it hangs on Waiting for Quorum.

Here's is my Corosync:

logging {
  debug: off
  to_syslog: yes
}

nodelist {
  node {
    name: ns524364
    nodeid: 1
    quorum_votes: 1
    ring0_addr: ns524364
  }

  node {
    name: ns545353
    nodeid: 4
    quorum_votes: 1
    ring0_addr: ns545353
  }

  node {
    name: ns541194
    nodeid: 3
    quorum_votes: 1
    ring0_addr: ns541194
  }

  node {
    name: ns535698
    nodeid: 2
    quorum_votes: 1
    ring0_addr: ns535698
  }

}

quorum {
  provider: corosync_votequorum
}

totem {
  cluster_name: mamae
  config_version: 62
  ip_version: ipv4
  secauth: on
  transport: udpu
  version: 2
  interface {
    ringnumber: 0
  }

}

Have you looked at:
https://pve.proxmox.com/wiki/Multic....29_instead_of_multicast.2C_if_all_else_fails

Do you have a mapping of hostnames to node ip in a DNS or each cluster nodes /etc/host file?
If not, use the "real" IPs for the nodes "ring0_addr" parameters.

 

[t.lamprecht]

The udpu transport needs a list of members. See this example.
https://linux.die.net/man/5/corosync.conf

You refer to outdated Corsync v1.X man pages and also not the official corosync repository:
https://github.com/corosync/corosync/blob/master/conf/corosync.conf.example.udpu

 

[masterdaweb]

Thank you for supporting me guys.

I figured it out using command "pvecm e 1" on the 4th node, and than I could successfully add to the cluster.

Many thanks.