Organization of a high-availability cluster

Clustering of server systems is one of the ways to ensure the reliability of data processing and storage systems, as well as constant provision of users with prompt and reliable information. With the clustering of server systems, a high level of availability is maintained. The article presents the construction of a high availability cluster. Several servers are used to increase the fault tolerance, which are connected through several switches. Several network interfaces are provided in servers using the technology of bonding. Also, servers provide a RAID array. Servers and switches are connected to uninterruptible power supplies. Uninterruptible power supplies are backed up. The software packages are presented with their configuration files. Software packages are necessary for the functioning of the cluster: bond, drbd, pacemaker, corosync. The pacemaker configuration and agent resource are described. Those agents was described: virtual IP address - for a cluster to have a single address, drbd for mirroring RAID arrays, programmControl for managing applications, wakeUp for controlling servers turning on and off. As a result, a high availability cluster was designed. It is reserved by the server part, network part and memory. Customer service is interrupted for a very short time when moving from one server to another.