HA cluster on 2 servers with ZFS

[asdrojd]

OS Virtual Environment 6.3-2
heve 2 edentical servers
3 hdd on each /dev/sda /dev/sdb /dev/sdc
/dev/sda for proxmox system
/dev/sdb /dev/sdc for zfs

on each was configured zfs mirror
root@host2:/etc/network# zpool status -v
pool: zfs
state: ONLINE
scan: none requested
config:

NAME STATE READ WRITE CKSUM
zfs ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
sdc ONLINE 0 0 0
sdd ONLINE 0 0 0

errors: No known data errors

was created zfs_pool for replication between servers
was created HA cluster, also HA group with both servers(restricted:yes nofailback:no)
created virtual server on node1(hdd on virtual server is on ZFS on node1)
virtual server from node1 was migrated on node2 (manually) -it works ok

next i want to test
node2 server was switched off
question how to start virtual server on node1 or automatically on manually
replication was several minutes ago and i undestood about loosing fresh information....

how to start virtual server now on node1 (in that time node2 is down, but i have replication on node1)
can not find anywhere
on node 1 i have(copy of) config of virtual server in /etc/pve/nodes/host2/qemu-server/100.conf
but i can't it copy to /etc/pve/nodes/host1/qemu-server

what to do next?

 

[mailinglists]

Ask your self, do you have quorum, with 2 node cluster, when one node dies?

 

[asdrojd]

Ask your self, do you have quorum, with 2 node cluster, when one node dies?

the question is what to do next?
ok automatic not working how to start manual (as i said before, i undestood about loosing info after replication) but i need to start this virtual server

 

[asdrojd]

after 2 days of searching the clear responce is
on live node start command
pvecm expected 1

and after several minutes virtual server will be automatically moved to live node and started
(if this virtual server previously was configured in HA->resources)

 

[aaron]

pvecm expected 1

Instead of playing around with those settings I highly suggest taking a look at the qdevice mechanism which can provide a third vote to the cluster.

See the documentation: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_corosync_external_vote_support

TL;DR: install the corosync-qnetd service on some machine that is not in the cluster and configure it in the cluster. It will give you a third vote and if one node is down you still hold 2 out of 3 and thus have the majority.