[SOLVED] Adding Node to Cluster Makes Other Nodes Un-pingable

[Keyninja]

Hi All. I'm trying to add a 16th node to our cluster, but when I try to add it, I get 1. locked out of the 16th node, and 2. It causes several other members of the cluster to become unpingable, as well as ALL nodes in the cluster showing as beig in Standalone mode, even though you can see them all in the list. Any ideas as to what's going on?

 

[oguz]

Hi.

Are all the nodes running the same PVE versions? They might not play well together if the versions are different.

Checking these files/commands might give some insight about the situation as well:
* /var/log/syslog
* pvecm status
* /etc/pve/.clusterlog

 

[Keyninja]

Hi.

Are all the nodes running the same PVE versions? They might not play well together if the versions are different.

Checking these files/commands might give some insight about the situation as well:
* /var/log/syslog
* pvecm status
* /etc/pve/.clusterlog

Thanks oguz! I'm guessing you mean in the currently active nodes, not node12?

 

[oguz]

Thanks oguz! I'm guessing you mean in the currently active nodes, not node12?

I think checking it on one or two nodes should be enough, especially since /etc/pve should have the same contents on all nodes (it's synced via cluster filesystem).

Would you mind telling how you're trying to add this node? Can you also check the PVE versions on the active and inactive nodes? `pveversion -v` should give you an output for the node you run it on.

EDIT: Now that I think of it, you should also check the hostnames of all the nodes in the cluster. It might be problematic if some nodes have the same hostname, causing problems in host resolution.

 

[Keyninja]

I think checking it on one or two nodes should be enough, especially since /etc/pve should have the same contents on all nodes (it's synced via cluster filesystem).

Would you mind telling how you're trying to add this node? Can you also check the PVE versions on the active and inactive nodes? `pveversion -v` should give you an output for the node you run it on.

EDIT: Now that I think of it, you should also check the hostnames of all the nodes in the cluster. It might be problematic if some nodes have the same hostname, causing problems in host resolution.

So, as far as their hostnames go, the schema is basically nodexx.yyyyyyyyy.com None of them should be the same. All of them have been installed with the latest 5.4 .iso, so their PVE version is all:

proxmox-ve: 5.4-1 (running kernel: 4.15.18-12-pve)
pve-manager: 5.4-3 (running version: 5.4-3/0a6eaa62)
pve-kernel-4.15: 5.3-3
pve-kernel-4.15.18-12-pve: 4.15.18-35
ceph: 12.2.11-pve1
corosync: 2.4.4-pve1
criu: 2.11.1-1~bpo90
glusterfs-client: 3.8.8-1
ksm-control-daemon: not correctly installed
libjs-extjs: 6.0.1-2
libpve-access-control: 5.1-8
libpve-apiclient-perl: 2.0-5
libpve-common-perl: 5.0-50
libpve-guest-common-perl: 2.0-20
libpve-http-server-perl: 2.0-13
libpve-storage-perl: 5.0-41
libqb0: 1.0.3-1~bpo9
lvm2: 2.02.168-pve6
lxc-pve: 3.1.0-3
lxcfs: 3.0.3-pve1
novnc-pve: 1.0.0-3
proxmox-widget-toolkit: 1.0-25
pve-cluster: 5.0-36
pve-container: 2.0-37
pve-docs: 5.4-2
pve-edk2-firmware: 1.20190312-1
pve-firewall: 3.0-19
pve-firmware: 2.0-6
pve-ha-manager: 2.0-9
pve-i18n: 1.1-4
pve-libspice-server1: 0.14.1-2
pve-qemu-kvm: 2.12.1-3
pve-xtermjs: 3.12.0-1
qemu-server: 5.0-50
smartmontools: 6.5+svn4324-1
spiceterm: 3.0-5
vncterm: 1.5-3
​

I'm not adding it fancily or anything, just using the web UI and copying the join information into node12's web UI "Join Cluster" dialog box. node12 comes up as the exact same version on all packages. And I have gone through all their host files and verified their public IP does indeed match up to their FQDN and hostname. None of them point to 12 or conflict with each other in any way.

 

[Keyninja]

So, now it poisoned node1 and I had to remove that as well. It was spitting a "could not resolve node1/node12" error. I tried re-installing node12 and totally changing the IP address, incremented the hostname up to node18, and added again, with the same results. So, I removed 1, and here we are. I figure this is probably corosync related, so here's my corosync.conf.

logging {
debug: off
to_syslog: yes
}

nodelist {
node {
name: node10
nodeid: 10
quorum_votes: 1
ring0_addr: 10.48.5.11
}
node {
name: node11
nodeid: 11
quorum_votes: 1
ring0_addr: 10.48.5.12
}
node {
name: node13
nodeid: 12
quorum_votes: 1
ring0_addr: 10.48.5.14
}
node {
name: node14
nodeid: 13
quorum_votes: 1
ring0_addr: 10.48.5.15
}
node {
name: node15
nodeid: 14
quorum_votes: 1
ring0_addr: 10.48.5.16
}
node {
name: node16
nodeid: 15
quorum_votes: 1
ring0_addr: 10.48.5.17
}
node {
name: node18
nodeid: 1
quorum_votes: 1
ring0_addr: 10.48.5.36
}
node {
name: node2
nodeid: 2
quorum_votes: 1
ring0_addr: 10.48.5.3
}
node {
name: node3
nodeid: 3
quorum_votes: 1
ring0_addr: 10.48.5.4
}
node {
name: node4
nodeid: 4
quorum_votes: 1
ring0_addr: 10.48.5.5
}
node {
name: node5
nodeid: 5
quorum_votes: 1
ring0_addr: 10.48.5.6
}
node {
name: node6
nodeid: 6
quorum_votes: 1
ring0_addr: 10.48.5.7
}
node {
name: node7
nodeid: 7
quorum_votes: 1
ring0_addr: 10.48.5.8
}
node {
name: node8
nodeid: 8
quorum_votes: 1
ring0_addr: 10.48.5.9
}
node {
name: node9
nodeid: 9
quorum_votes: 1
ring0_addr: 10.48.5.10
}
}

quorum {
provider: corosync_votequorum
}

totem {
cluster_name: hou1-vpc1
config_version: 25
interface {
bindnetaddr: 10.48.5.2
ringnumber: 0
}
ip_version: ipv4
secauth: on
version: 2
}
​

All IP's are non-routable privates.

 

[Keyninja]

I ended up reinstalling the cluster, it was a DNS issue. Bleh.