dmraid tool design document v1.0.0-rc5f Heinz Mauelshagen 2004.11.05
----------------------------------------------------------------------------
The dmraid tool supports RAID devices (RDs) such as ATARAID with
device-mapper (dm) in Linux 2.6 avoiding the need to install a
vendor specific (binary) driver to access them.
It supports multiple on-disk RAID metadata formats and is open for
extension with new ones.
dmraid supports activation of RDs and *updates* of the ondisk metadata
(eg, to record disk failures) for the Intel Software RAID (IMSM) format.
Others may follow.
It can optionally erase on disk metadata in case of multiple format
signatures on a device.
See future enhancements at the end.
Functional requirements:
------------------------
1. dmraid must be able to read multiple vendor specific ondisk
RAID metadata formats:
o ATARAID
- Adaptec HostRAID ASR
- Highpoint 37x/45x
- Intel Software RAID
- JMicron ATARAI
- LSI Logic MegaRaid
- NVidia RAID
- Promise FastTrak
- Silicon Image Medley
- SNIA DDF1
- VIA Software RAID
- DOS partitions on SW RAIDs
2. dmraid shall be open to future extensions by other ondisk RAID formats.
3. dmraid shall generate the necessary dm table(s) defining
the needed mappings to address the RAID data.
4. Device discovery, activation, deactivation and
property display shall be supported.
5. Spanning of disks, RAID0, RAID1 and RAID10 shall be supported
(in order to be able to support SNIA DDF, higher raid levels need
implementing in form of respective dm targets; eg, RAID5);
Some vendors do have support for RAID5 already which is outside the scope
of dmraid because of the lag of a RAID5 target in device-mapper!
6. Activation of MSDOS partitions in RAID sets shall be supported.
Feature set definition:
-----------------------
Feature set summarizes as: Discover, Activate, Deactivate, Display.
o Discover (1-n RD)
1 scan active disk devices identifying RD.
2 try to find an RD signature and if recognized,
add the device to the list of RDs found.
3 Abstract the metadata describing the RD layout and translate the vendor
specific format into it.
o Activate (1-n RD)
1 group devices into abstracted RAID sets (RS) conforming to
their respective layout (SPAN, RAID0, RAID1, RAID10).
2 generate dm mapping tables for a/those RS(s) derived from the abstracted
information about RS(s) and RD(s).
3 create multiple/a dm device(s) for each RS to activate and
load the generated table(s) into the device.
4 Recusively discover and activate MSDOS partitions in active RAID sets.
o Deactivate (1-n RD)
1 remove the dm device(s) making up an RS; can be a stacked hierachy of
devices (eg, RAID10: RAID1 on top of n RAID0 devices).
o Display (1-n RD)
1 display RS and RD properties
(eg, display information kept with RS(s) such as size and type)
Technical specification:
------------------------
All functions returning int or a pointer return 0/NULL on failure.
o RAID metadata format handler
Tool calls the following function to register a vendor specific
format handler; in case of success, a new instance with methods is
accessible to the high level metadata handling functions (see below):
- int register_format_handler(struct lib_context *lc,
struct dmraid_format *fmt);
x returns !0 on successfull format handler registration
- Format handler methods:
x struct raid_dev *(read)(struct lib_context *lc, struct dev_info* di);
- returns 'struct raid_dev *' describing the RD (eg, offset, length)
x int (*write)(struct lib_context *lc, struct raid_dev* rd, int erase);
- returns != 0 successfully written the vendor specific metadata
back to the respective RD rd (optionally erasing the
metadata area(s) in case erase != 0)
x struct raid_set (*group)(struct lib_context *lc,
struct raid_dev *raid_dev)
- returns pointer to RAID set structure on success
x int (*check)(struct lib_context *lc, struct raid_set *raid_set)
- returns != 0 in case raid set is consistent
x void (*log)(struct lib_context *lc, struct raid_dev *rd)
- display metadata in native format (ie. all vendor specific
metadata fields) for RD rd
o Discover
1 retrieve block device information from sysfs for all disk
devices by scanning /SYSFS_MOUNTPOINT/block/[sh]d*;
keep information about the device path and size which is the base
to find the RAID signature on the device in a linked list
of type 'struct dev_info *'.
2 walk the list and try to read RD metadata signature off the device
trying vendor specific read methods (eg, Highpoint...) in turn;
library exposes interface to register format handlers for vendor
specific RAID formats in order to be open for future extensions
(see register_format_handler() above).
Tool calls the following high level function which hides
the iteration through all the registered format handler methods:
x void discover_devices(struct lib_context *lc)
- returns != 0 in case it discovered all supported disks (IDE, SCSI)
and added them to a list
x void discover_raid_devices(struct lib_context *lc,
char **fmt_names,
char **devices);
- discovers all suported RDs using the list filled by
discover_devices() and adds them to a list
x void discover_partitions(struct lib_context *lc);
- discovers all MSDOS partitions in active RAID sets and builds
RAID sets with one linear device hanging off for activation.
x int count_devices(struct lib_context *lc, enum dev_type type);
- returns number of devices found by discover_devices() and
discover_raid_devices() of specified type (eg, DEVICE, RAID, ...)
x int perform(struct lib_context *lc, char **argv);
- returns != 0 on success performing various actions
(ie. activation/deacivation of RAID sets, displaying properties, ...)
o Activate 1
Tool calls the following high level function which hide
the details of the RAID set assembly:
x int group_set(struct lib_context *lc, char *name);
- returns != 0 on successfully grouping an RS/RSs
o Activate 2+3
- don't activate non-RAID1 devices which have an invalid set check result
and display an error
- create the ASCII dm mapping table by iterating through the list
of RD in a particular set, retrieving the layout (SPAN, ...)
the device path, the offset into the device and the length to map
and the stripe size in case of RAID
- create a unique device_name
- call device-mapper library to create the mapped device and load
the mapping table using this function:
x int activate_set(struct lib_context *lc, void *rs)
- returns != 0 in case of successfull RAID set activation
o Activate 4
- activate MSDOS partitioned RAID sets as in 2+3
o Deactivate
- check if a partitioned RAID set is active and deactivate it first
- check if the RAID set is active and call device-mapper library to
remove the mapped device (recursively in case of a mapped-device hierarchy)
using this function:
x int deactivate_set(struct lib_context *lc, void *rs)
- returns != 0 in case of successfull RAID set deactivation
o Display
- list all block devices found
- list all (in)active RD
- display properties of a particular/all RD devices
(eg, members of the set by block device name and offset/length mapped
to those...)
x void display_devices(struct lib_context *lc, enum dev_type type);
- display devices of dev_type 'type' (eg, RAID, DEVICE, ...)
x void display_set(struct lib_context *lc, void *rs,
enum active_type type, int top);
- display RS of active_type 'type' (ACTIVE, INACTIVE, ALL)
Code directory tree:
--------------------
.
|-- autoconf
|-- doc
|-- include
| `-- dmraid
|-- lib
| |-- activate
| |-- datastruct
| |-- device
| |-- display
| |-- events
| |-- format
| | |-- ataraid
| | |-- ddf
| | |-- partition
| | `-- template
| |-- locking
| |-- log
| |-- metadata
| |-- misc
| |-- mm
| `-- register
|-- logwatch
|-- man
`-- tools
24 directories
Enhancements coming (initially with the Intel Software RAID):
-------------------------------------------------------------
o enhance write support to update ondisk metadata
- to restore metadata backups
- to record disk failures
o support to log state (eg, sector failures) in standard/vendor ondisk logs;
needs above write support
o status daemon to monitor RAID set sanity
(eg, disk failure, hot spare rebuild, ...) and
frontend with CLI
Open questions:
---------------
o we need to prioritize on device-mapper targets for higher RAID levels
(in particular we'ld need RAID4+5 to support some ATARAID formats)
