Splynx High Availability setup
Splynx High Availability setup allows you to use Splynx in Virtual Machine on a group of physical servers. We will call this group of servers - ‘cluster’ and every server we will call - ‘node’
We use Proxmox Virtual Environment with Ceph in this tutorial.
- at least three servers
- two storage drives on each server (one for system and one for Ceph)
- all servers must be in the same network
- two network interface controllers for each server (one for Ceph)
Figure 1. Network diagram
Install Proxmox to all nodes
Download installation ISO from https://www.proxmox.com/en/downloads/category/iso-images-pve
Burn it to CD or create bootable Flash USB drive
Hostname and IP Address must be different on different nodes
After installation – remove Proxmox CD (Flash drive)
Disable the enterprise repository that is configured by default, add the no-subscription repository.
#deb https://enterprise.proxmox.com/debian/pve stretch pve-enterprise deb http://download.proxmox.com/debian stretch pve-no-subscription
Run in shell (on each node):
nodeX> apt-get update && apt-get -y dist-upgrade
Reboot nodes after upgrade.
We must merge all nodes into one cluster. Make sure that each node is installed with the final hostname and IP configuration. Changing the hostname and IP is not possible after cluster creation.
Run in shell on the first node:
node1> pvecm create splynx-cluster</pre>
where ‘splynx-cluster’ – is the name of cluster. Can be any.
Add the rest nodes to the cluster. On every node run:
node2> pvecm add IP-ADDRESS-OF-NODE1
node3> pvecm add IP-ADDRESS-OF-NODE1
Ceph is a network-based file system. Ceph replicates data and makes it fault-tolerant. Your data will be in safe even if one (or more) servers will fail. This articles describes how to setup and run Ceph storage services directly on Proxmox VE nodes.
Install Ceph to each node:
node1> pveceph install
node2> pveceph install
node3> pveceph install
We use separated NICs and sepatated IP network for Ceph traffic. Part of /etc/network/interfaces:
auto ens19 iface ens19 inet static address 172.30.250.1 #(172.30.250.2 – on 2nd node, 172.30.250.3 – on 3rd node) netmask 255.255.255.0
After editing /etc/network/interfaces - reboot node to apply changes.
Create an initial Ceph configuration on just one node:
node1> pveceph init --network 172.30.250.0/24
This creates an initial config in /etc/pve/ceph.conf. This file is automatically distributed to all Proxmox VE nodes.
If you get error message like this:
unable to open file '/etc/pve/ceph.conf.tmp.12688' - Permission denied
Reboot your nodes and try again.
Create Ceph monitors on all nodes:
node1> pveceph createmon
node2> pveceph createmon
node3> pveceph createmon
Erase partition table of Ceph drive(s) and create OSD(s) on it. Run on each node:
nodeX> ceph-disk zap /dev/sdb nodeX> pveceph createosd /dev/sdb
*We use /dev/sda for system and /dev/sdb for Ceph in this tutorial
Create Ceph pool.
Run on one node:
node3> pveceph createpool default-pool -add_storages true</pre>
This creates pool with name ‘default-pool’ and adds storages for VMs and containers to it. Pool name can be any.
Open your web-browser and type https://IP-OF-ANY-NODE:8006. If you get certificate error – just ignore it (add to exceptions).
User name: root
Password: the password you have entered during installation.
We will install Ubuntu-16.04-server. Download ISO image to PC. Then upload it from PC to local storage of one of Proxmox nodes:
Create VM on the node:
Use Ceph storage for VM:
Start VM, open Console and install Linux as usual:
Remove ISO from VM:
Let’s find IP address of VM. Open Console, log-in and type
We can see that VM has IP address 192.168.77.246
Connect to VM using SSH (we found IP address in the previous step)
sudo apt-get update && sudo apt-get -y dist-upgrade
wget -qO- https://deb.splynx.com/setup | sudo bash - sudo apt-get install splynx
Create HA group:
Add VM to this group:
Now we have running VM on node1. If node1 fails – ProxmoxVE automatically restarts VM on the other node in about 2 minutes.
You can also initiate VM migration to other node:
bash> ha-manager migrate vm:100 DESTINATION-NODE
This uses online migration and tries to keep the VM running. Online migration needs to transfer all used memory over the network, so it is sometimes faster to stop VM, then restart it on the new node. This can be done using the relocate command:
bash> ha-manager relocate vm:100 DESTINATION-NODE</pre>