Automatic startup of a cluster system after occurrence of a recoverable error
View Patent ↗The invention relates to a method for the automatic startup of a cluster ( 10 ) after an error has occurred in a node ( 12, 14 ) of said cluster ( 10 ) that led to a reboot of the node ( 12, 14 ). The inventive method is characterized in that it automatically recognizes whether the error can be recovered and the cluster ( 10 ) can be automatically started up. The inventive method allows for the automatic return of the cluster ( 10 ) to its operation state after occurrence of an error, thereby reducing down-times of the system.
1. A method of starting up a cluster after an error encompassing a number of nodes in which, in the event of an error, a second node takes over jobs of a first node, comprising:
determining at least one first and one second error class, where the first error class includes an automatic startup of the cluster;
analyzing the error that led to a reboot of the first node;
classifying the error into one of the error classes; and
automatically starting-up the cluster with the nodes if the error has been classified into the first error class.
2. The method in accordance with claim 1 , wherein the cluster is not started up automatically for the second error class.
3. The method in accordance with claim 1 , wherein the analysis of the error includes at least power-on detection by reading out and comparing the lifetimes of the nodes.
4. The method in accordance with claim 1 , wherein the cluster is a SUN cluster.
5. The method in accordance with claim 1 , wherein the startup of the cluster after the reboot of the first node is undertaken when the first node has reached a state in which it has full functionality available to it, in which the network processes are running and file systems are mounted.
6. The method in accordance with claim 1 , wherein the cluster is an active-active cluster.
7. The method in accordance with claim 1 , wherein the cluster is an active-passive cluster.
8. The method in accordance with claim 1 , wherein analysis of the error for the reboot is undertaken at two nodes assigned to each other for the cluster.
9. The method in accordance with claim 1 , wherein the startup of the cluster is performed with coordinated nodes, whereby the nodes are included sequentially into the cluster each with different commands and whereby each of these inclusions is checked to ensure that it is free from errors.
10. The method in accordance with claim 6 , wherein the coordination of the nodes is undertaken by determining a state of the nodes on both sides of the cluster at pre-specified coordination times.
11. A computer program product stored in memory, comprising software code sections executable by at least one processor, comprising:
determining at least one first and one second error class, where the first error class includes an automatic startup of the cluster;
analyzing the error that led to a reboot of the first node;
classifying the error into one of the error classes; and
automatically starting-up the cluster with the nodes if the error has been classified into the first error class.
12. A cluster stored on a computer, the cluster performing:
determining at least one first and one second error class, where the first error class includes an automatic startup of the cluster;
analyzing the error that led to a reboot of the first node;
classifying the error into one of the error classes; and
automatically starting-up the cluster with the nodes if the error has been classified into the first error class.
13. A device, comprising at least one unit to execute the following:
determining at least one first and one second error class, where the first error class includes an automatic startup of the cluster;
analyzing the error that led to a reboot of the first node;
classifying the error into one of the error classes; and
automatically starting-up the cluster with the nodes if the error has been classified into the first error class.