Cluster Installation

Since all services are available multiple times in a cluster network, the cluster network not only provides fail-safe operation but also improves performance, since the tasks at hand can be distributed among the services automatically. Failures of individual services can thus be compensated in most cases.

A good option for supplementing a IQ4docs cluster is load balancing via DNS. To accomplish this, several IP addresses are specified for a (cluster) name in the DNS, which are then resolved differently for each query. In this way, for example, the WebAdmin could always be called with the same hostname, even if a cluster node is not accessible. If the name is just resolved with the IP address leading to the failed cluster node, the request must be executed again to be routed to another (working) cluster node.

In a cluster installation, there are certain requirements that are mandatory for a functioning system.

• If a cluster node has to be shut down or restarted for updates or similar, make sure that the remaining cluster nodes continue to run. A restart of the cluster servers must always be performed with a time delay – never simultaneously. After the restart, wait until all services can once again be reached (especially RabbitMQ and MongoDB).

• The network connection between the cluster nodes must not be interrupted – the primary cluster node cannot distinguish whether the server is no longer accessible because the network is no longer functioning or because the server has been shut down. At least two cluster nodes must still be able to reach each other. If this cannot be ensured, we recommend stopping the services beforehand and restarting them afterwards.

• With Windows updates as well, it is important to make sure that these are not carried out simultaneously, but at different times. In many cases, a restart of the operating system must be initiated because of the update.

• In the case of virtual servers, even stopping a server for a short time (e.g., by suspending the virtual machine) can result in cluster nodes no longer being connected correctly. The effects are similar to a network interruption.

• Short interruptions can already be caused, for example, by the creation of snapshots or by a very high load. By failing to respond to keep-alive packets, the effects are similar to those of a network interruption.

• It may be necessary to make adjustments to the device settings.

For correct operation of the devices of a IQ4docs cluster, please note that device settings should also be made appropriately, see Add devices.

To ensure that the files in the directory system (print and scan files and other files used by workflows) can be accessed by all servers, they must be stored on fail-safe Shared Storage media. Files in the directory system are not automatically replicated by IQ4docs. Create a folder on a shared storage medium and make sure that the services of all servers can access it, see also Storage locations.

The Shared Storage must be located separately from the IQ4docs servers and be designed to be fail-safe or highly available.