What is “Network Ping Test IP”?

Upon creation of a cluster there is an option to enter network ping test ip...

This IP is used by the server to determine if the network is still available. Cluster nodes will ping this address constantly, if none of the IPs is available. The node will know that it is offline and will give its services to the online node

It is recommended to put more than just one IP.

This can be any device on your network that is constantly on, router, gateway, printer, mail server etc.. ..

What does HA Cluster?

HA Storage Clusters are used to provide critical shared data on a network, and as system and data drives for physical and virtual production servers and critical workstations.

They operate by using redundant servers that provide continued service even when individual system components or network elements fail.

Without clustering, if a storage server is down, the data and all related applications and services will be unavailable until the server is back online.

HA Clustering remedies this situation by detecting hardware and software faults, and immediately sending storage requests to the other server in the cluster without requiring administrative intervention.

This benefit is not only for the failures and crashes that will inevitably happen, but also for planned maintenance. A server in the cluster may be taken offline for maintenance and brought back online without interrupting your users or customers.

What are hardware recommendations for HA Cluster?

Installation Disk

For best performance the euroNAS cluster software should be installed on an SSD disk with a minimum size of 40 GB

System RAM

The server should have at least 8 GB of RAM

Networking

There must be one or more reliable IP Addresses on the Corporate Network that both servers can reach for proper operation of the High-Availability/Custer features. An example would be the router which is on the Corporate Network or connects the servers to the Corporate Network.

For best performance and high availability the server should have 4 individual Network Interface Cards with a minimum bandwidth of 1 Gigabit.

They should be configured as two separate bonded networks. One of the bonds should be connected to the Corporate Network. The other bond should be connected directly to the other server on a private Replication Network. If the servers are in the same location they should be cabled directly to each other without any switches between them if possible.

For top performance the Replication Network should be 10 Gigabit.

If dual-port cards are used each port on the card should be assigned to a separate bond so that the failure of one card does not break the entire bond.

Card 1 Port 1 -> bond0 (Corporate Network)
Card 1 Port 2 -> bond1 (Replication Network)
Card 2 Port 1 -> bond0 (Corporate Network)
Card 2 Port 2 -> bond1 (Replication Network)

Quad-port cards should be avoided

DHCP should not be used unless static leases are provided. Using DHCP without static leases will cause the cluster to fail when the IP addresses change.

Is vSphere High Availability (HA) supported by HA Cluster?

euroNAS HA Cluster is perfect match for vSphere High Availability (HA) functionality.

While vSphere High Availability (HA) takes care about the virtual machines – euroNAS HA Cluster makes sure that the storage is always available.

Should both servers have identical hardware?

Both servers should have same disks because they will be replicated in real time.

It is not nessesary to have other hardware identical, however it should be kept in mind that in case one server fails the other will take over. So both servers should be capable of handling the server load.

What is “cluster drive”?

Cluster drive represents a mirror of 2 drives on each inidividual server.

This 2 drives can be a single disk, software or raid array. They are mirrored in real-time. On this drives shares and iSCSI Targets are installed.

When should I use HA Cluster?

euroNAS HA Cluster Software interlinks your data servers with redundant servers that provide continued service, even when individual system components or network elements fail. It supports NFS and ISCSi protocols.

High Availability Clusters are especially designed for:

  • network systems with critical shared data
  • all kinds of virtualization
  • system and data drives for physical and virtual production servers
  • critical workstations

Increasing the disk timeout values under Linux

euroNAS HA Cluster provides you the highest level of redundancy. It even protects you if one of the replication servers fail. In this case all ressources will be moved to another running server automatically.

During failover the ressources are unavailable during the failover period. In most cases OS can handle this but sometimes it is nessesary to change default timeout settings.

When connecting to Linux using iSCSI timings depend on 2 factors:

  • Timeout of the disk and
  • of the open-iscsi initiator.

You can see the current time out values of your iSCSI disk with following command:
find /sys/class/scsi_generic/*/device/timeout -exec grep -H . ‘{}’ \;

Value can be changed by editing timeout settings in /sys – for example:
echo 180 > /sys/block/sdd/device/timeout

This setting is not remaining after reboot so putting it into a start script is recommended.