| | 1 | = Introduction = |
| | 2 | The purposes of integrate Lustre, DRBD and heartbeat are as follows: |
| | 3 | == 1. For data replication == |
| | 4 | Lustre not only provides high-performance I/O but also is capable of aggregating any pieces of storages in cluster environment. The only fly in the ointment is the lack of data replication, and DRBD is used to make up this shortcomings. |
| | 5 | DRBD is a well-known tools for High-Availability in cluster. |
| | 6 | == 2. For Auto-Detection of Failures == |
| | 7 | Heartbeat is capable to detect the failures of pre-configure node. |
| | 8 | [[BR]] |
| | 9 | This solution is exactly what we need in a High-Availability cluster environment. |
| | 10 | The ideal work flow is as follows:[[BR]] |
| | 11 | Working --> write with replication data --> detect failures occur --> client switch to backup node to access data continuously --> failure have been fixed up --> client switch to original node to access data continuously. |
| | 12 | [[BR]] |
| | 13 | Lustre + DRBD + Heartbeat can almost achieve this goal. |
| | 14 | = Requirements = |
| | 15 | Requirement is pretty simple, just need lustre, drbd and Heartbeat. DRBD integrate with lustre is more complicated than heartbeat, since drbd is a kernel module as lustre. [[BR]] |
| | 16 | = Testing Steps = |
| | 17 | == Install Lustre == |
| | 18 | Please refer to [wiki: xxxx Lustre installation guide] |
| | 19 | == Install DRBD == |
| | 20 | |
| | 21 | == Install Heartbeat == |
| | 22 | |
| | 23 | = Conclusion = |
