[SOLVED] cluster join failed: 500 Can't connect to ... (hostname verification failed)

[cglmicro]

Hi guys.

I'm trying to add a new slave (pmg12) on my 3 nodes cluster to connect to my master (pmg10 with the IP 198.xxx.xxx.xxx), and I'm getting this error when trying to add it in CLI, and similar error from GUI (the xxx are replaced with my ip of course):

(from pmg12)
pmgcm join 198.xxx.xxx.xxx
cluster join failed: 500 Can't connect to 198.xxx.xxx.xxx:8006 (hostname verification failed)

and from the gui perspective I see:

TASK ERROR: 401 permission denied - invalid PMG ticket

All my nodes are version 7.1.1, freshly rebooted my master and the new PMG I want to add.
I'm able to ping and to "ssh 198.xxx.xxx.xxx".
I'm able to access the GUI of both PMG (so port 8006 is working fine).
I tried to disable the firewall on the master and on the new PMG.
The timezone are the same on both servers.
On my new PMG12 two services are not running (pmgtunnel and pmgmiror), is it because they aren't required until a cluster is joined?

Where should I start to look?

Thank you.

 

[Stoiko Ivanov]

a) is the time synchronized between the nodes (is NTP running and synchronised)?
b) please try joining on the command-line by pasting the output of `pmgcm join-cmd` (run on the master) - to the fresh node
c) on a hunch - have you enabled 2FA for your root-user?

On my new PMG12 two services are not running (pmgtunnel and pmgmiror), is it because they aren't required until a cluster is joined?

yes - see `systemctl cat pmgmirror`:

[Unit]
..snip..
ConditionPathExists=/etc/pmg/cluster.conf
..snip..

I hope this helps!

 

[cglmicro]

Congrats Stoiko, it was the 2FA I enabled.
I've disabled it, joined the cluster, and enabled it again.

I had a similar bug in PBS or PVE, can't remember.

Thanks a lot !!

 

[Stoiko Ivanov]

I had a similar bug in PBS or PVE, can't remember.

cluster-join would speak more for PVE - but glad we found it and it worked now