| Philipp Reisner | b411b36 | 2009-09-25 16:07:19 -0700 | [diff] [blame] | 1 | # |
| 2 | # DRBD device driver configuration |
| 3 | # |
| 4 | |
| 5 | comment "DRBD disabled because PROC_FS, INET or CONNECTOR not selected" |
| Johannes Thoma | 89f01d5 | 2009-12-29 16:38:04 +0100 | [diff] [blame] | 6 | depends on PROC_FS='n' || INET='n' || CONNECTOR='n' |
| Philipp Reisner | b411b36 | 2009-09-25 16:07:19 -0700 | [diff] [blame] | 7 | |
| 8 | config BLK_DEV_DRBD |
| 9 | tristate "DRBD Distributed Replicated Block Device support" |
| 10 | depends on PROC_FS && INET && CONNECTOR |
| 11 | select LRU_CACHE |
| 12 | default n |
| 13 | help |
| 14 | |
| 15 | NOTE: In order to authenticate connections you have to select |
| 16 | CRYPTO_HMAC and a hash function as well. |
| 17 | |
| 18 | DRBD is a shared-nothing, synchronously replicated block device. It |
| 19 | is designed to serve as a building block for high availability |
| 20 | clusters and in this context, is a "drop-in" replacement for shared |
| 21 | storage. Simplistically, you could see it as a network RAID 1. |
| 22 | |
| 23 | Each minor device has a role, which can be 'primary' or 'secondary'. |
| 24 | On the node with the primary device the application is supposed to |
| 25 | run and to access the device (/dev/drbdX). Every write is sent to |
| 26 | the local 'lower level block device' and, across the network, to the |
| 27 | node with the device in 'secondary' state. The secondary device |
| 28 | simply writes the data to its lower level block device. |
| 29 | |
| 30 | DRBD can also be used in dual-Primary mode (device writable on both |
| 31 | nodes), which means it can exhibit shared disk semantics in a |
| 32 | shared-nothing cluster. Needless to say, on top of dual-Primary |
| 33 | DRBD utilizing a cluster file system is necessary to maintain for |
| 34 | cache coherency. |
| 35 | |
| 36 | For automatic failover you need a cluster manager (e.g. heartbeat). |
| 37 | See also: http://www.drbd.org/, http://www.linux-ha.org |
| 38 | |
| 39 | If unsure, say N. |
| 40 | |
| Philipp Reisner | b411b36 | 2009-09-25 16:07:19 -0700 | [diff] [blame] | 41 | config DRBD_FAULT_INJECTION |
| 42 | bool "DRBD fault injection" |
| 43 | depends on BLK_DEV_DRBD |
| 44 | help |
| 45 | |
| 46 | Say Y here if you want to simulate IO errors, in order to test DRBD's |
| 47 | behavior. |
| 48 | |
| 49 | The actual simulation of IO errors is done by writing 3 values to |
| 50 | /sys/module/drbd/parameters/ |
| 51 | |
| 52 | enable_faults: bitmask of... |
| 53 | 1 meta data write |
| 54 | 2 read |
| 55 | 4 resync data write |
| 56 | 8 read |
| 57 | 16 data write |
| 58 | 32 data read |
| 59 | 64 read ahead |
| 60 | 128 kmalloc of bitmap |
| 61 | 256 allocation of EE (epoch_entries) |
| 62 | |
| 63 | fault_devs: bitmask of minor numbers |
| 64 | fault_rate: frequency in percent |
| 65 | |
| 66 | Example: Simulate data write errors on /dev/drbd0 with a probability of 5%. |
| 67 | echo 16 > /sys/module/drbd/parameters/enable_faults |
| 68 | echo 1 > /sys/module/drbd/parameters/fault_devs |
| 69 | echo 5 > /sys/module/drbd/parameters/fault_rate |
| 70 | |
| 71 | If unsure, say N. |