2004-06-12 21:16:10 +00:00
|
|
|
/*-
|
2017-11-27 15:17:37 +00:00
|
|
|
* SPDX-License-Identifier: BSD-2-Clause-FreeBSD
|
|
|
|
*
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
* Copyright (c) 2004, 2005, 2007 Lukas Ertl
|
2004-06-12 21:16:10 +00:00
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
|
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
|
|
* SUCH DAMAGE.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
2011-07-11 05:22:31 +00:00
|
|
|
#include <sys/types.h>
|
2008-10-01 14:50:36 +00:00
|
|
|
#include <sys/endian.h>
|
2004-06-12 21:16:10 +00:00
|
|
|
#include <sys/malloc.h>
|
2011-07-11 05:22:31 +00:00
|
|
|
#include <sys/sbuf.h>
|
2004-06-12 21:16:10 +00:00
|
|
|
#include <sys/systm.h>
|
|
|
|
|
|
|
|
#include <geom/geom.h>
|
2019-08-07 19:28:35 +00:00
|
|
|
#include <geom/geom_dbg.h>
|
2004-06-12 21:16:10 +00:00
|
|
|
#include <geom/vinum/geom_vinum_var.h>
|
|
|
|
#include <geom/vinum/geom_vinum.h>
|
|
|
|
|
2008-10-01 14:50:36 +00:00
|
|
|
#define GV_LEGACY_I386 0
|
|
|
|
#define GV_LEGACY_AMD64 1
|
|
|
|
#define GV_LEGACY_SPARC64 2
|
|
|
|
#define GV_LEGACY_POWERPC 3
|
|
|
|
|
|
|
|
static int gv_legacy_header_type(uint8_t *, int);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Here are the "offset (size)" for the various struct gv_hdr fields,
|
|
|
|
* for the legacy i386 (or 32-bit powerpc), legacy amd64 (or sparc64), and
|
|
|
|
* current (cpu & endian agnostic) versions of the on-disk format of the vinum
|
|
|
|
* header structure:
|
|
|
|
*
|
|
|
|
* i386 amd64 current field
|
|
|
|
* -------- -------- -------- -----
|
|
|
|
* 0 ( 8) 0 ( 8) 0 ( 8) magic
|
|
|
|
* 8 ( 4) 8 ( 8) 8 ( 8) config_length
|
|
|
|
* 12 (32) 16 (32) 16 (32) label.sysname
|
|
|
|
* 44 (32) 48 (32) 48 (32) label.name
|
|
|
|
* 76 ( 4) 80 ( 8) 80 ( 8) label.date_of_birth.tv_sec
|
|
|
|
* 80 ( 4) 88 ( 8) 88 ( 8) label.date_of_birth.tv_usec
|
|
|
|
* 84 ( 4) 96 ( 8) 96 ( 8) label.last_update.tv_sec
|
|
|
|
* 88 ( 4) 104 ( 8) 104 ( 8) label.last_update.tv_usec
|
|
|
|
* 92 ( 8) 112 ( 8) 112 ( 8) label.drive_size
|
|
|
|
* ======== ======== ========
|
|
|
|
* 100 120 120 total size
|
|
|
|
*
|
|
|
|
* NOTE: i386 and amd64 formats are stored as little-endian; the current
|
|
|
|
* format uses big-endian (network order).
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Checks for legacy format depending on platform. */
|
|
|
|
static int
|
|
|
|
gv_legacy_header_type(uint8_t *hdr, int bigendian)
|
|
|
|
{
|
|
|
|
uint32_t *i32;
|
|
|
|
int arch_32, arch_64, i;
|
|
|
|
|
2016-04-29 20:56:58 +00:00
|
|
|
/* Set arch according to endianness. */
|
2008-10-01 14:50:36 +00:00
|
|
|
if (bigendian) {
|
|
|
|
arch_32 = GV_LEGACY_POWERPC;
|
|
|
|
arch_64 = GV_LEGACY_SPARC64;
|
|
|
|
} else {
|
|
|
|
arch_32 = GV_LEGACY_I386;
|
|
|
|
arch_64 = GV_LEGACY_AMD64;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* if non-empty hostname overlaps 64-bit config_length */
|
|
|
|
i32 = (uint32_t *)(hdr + 12);
|
|
|
|
if (*i32 != 0)
|
|
|
|
return (arch_32);
|
|
|
|
/* check for non-empty hostname */
|
|
|
|
if (hdr[16] != 0)
|
|
|
|
return (arch_64);
|
|
|
|
/* check bytes past 32-bit structure */
|
|
|
|
for (i = 100; i < 120; i++)
|
|
|
|
if (hdr[i] != 0)
|
|
|
|
return (arch_32);
|
|
|
|
/* check for overlapping timestamp */
|
|
|
|
i32 = (uint32_t *)(hdr + 84);
|
|
|
|
|
|
|
|
if (*i32 == 0)
|
|
|
|
return (arch_64);
|
|
|
|
return (arch_32);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Read the header while taking magic number into account, and write it to
|
|
|
|
* destination pointer.
|
|
|
|
*/
|
|
|
|
int
|
|
|
|
gv_read_header(struct g_consumer *cp, struct gv_hdr *m_hdr)
|
|
|
|
{
|
|
|
|
struct g_provider *pp;
|
|
|
|
uint64_t magic_machdep;
|
|
|
|
uint8_t *d_hdr;
|
|
|
|
int be, off;
|
|
|
|
|
|
|
|
#define GV_GET32(endian) \
|
|
|
|
endian##32toh(*((uint32_t *)&d_hdr[off])); \
|
|
|
|
off += 4
|
|
|
|
#define GV_GET64(endian) \
|
|
|
|
endian##64toh(*((uint64_t *)&d_hdr[off])); \
|
|
|
|
off += 8
|
|
|
|
|
|
|
|
KASSERT(m_hdr != NULL, ("gv_read_header: null m_hdr"));
|
|
|
|
KASSERT(cp != NULL, ("gv_read_header: null cp"));
|
|
|
|
pp = cp->provider;
|
|
|
|
KASSERT(pp != NULL, ("gv_read_header: null pp"));
|
|
|
|
|
2011-05-25 11:14:26 +00:00
|
|
|
if ((GV_HDR_OFFSET % pp->sectorsize) != 0 ||
|
|
|
|
(GV_HDR_LEN % pp->sectorsize) != 0)
|
|
|
|
return (ENODEV);
|
|
|
|
|
2008-10-01 14:50:36 +00:00
|
|
|
d_hdr = g_read_data(cp, GV_HDR_OFFSET, pp->sectorsize, NULL);
|
|
|
|
if (d_hdr == NULL)
|
|
|
|
return (-1);
|
|
|
|
off = 0;
|
|
|
|
m_hdr->magic = GV_GET64(be);
|
|
|
|
magic_machdep = *((uint64_t *)&d_hdr[0]);
|
|
|
|
/*
|
|
|
|
* The big endian machines will have a reverse of GV_OLD_MAGIC, so we
|
|
|
|
* need to decide if we are running on a big endian machine as well as
|
|
|
|
* checking the magic against the reverse of GV_OLD_MAGIC.
|
|
|
|
*/
|
|
|
|
be = (m_hdr->magic == magic_machdep);
|
|
|
|
if (m_hdr->magic == GV_MAGIC) {
|
|
|
|
m_hdr->config_length = GV_GET64(be);
|
|
|
|
off = 16;
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.sysname, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.name, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
m_hdr->label.date_of_birth.tv_sec = GV_GET64(be);
|
|
|
|
m_hdr->label.date_of_birth.tv_usec = GV_GET64(be);
|
|
|
|
m_hdr->label.last_update.tv_sec = GV_GET64(be);
|
|
|
|
m_hdr->label.last_update.tv_usec = GV_GET64(be);
|
|
|
|
m_hdr->label.drive_size = GV_GET64(be);
|
|
|
|
} else if (m_hdr->magic != GV_OLD_MAGIC &&
|
|
|
|
m_hdr->magic != le64toh(GV_OLD_MAGIC)) {
|
|
|
|
/* Not a gvinum drive. */
|
|
|
|
g_free(d_hdr);
|
|
|
|
return (-1);
|
|
|
|
} else if (gv_legacy_header_type(d_hdr, be) == GV_LEGACY_SPARC64) {
|
2008-10-26 17:20:37 +00:00
|
|
|
G_VINUM_DEBUG(1, "detected legacy sparc64 header");
|
2008-10-01 14:50:36 +00:00
|
|
|
m_hdr->magic = GV_MAGIC;
|
|
|
|
/* Legacy sparc64 on-disk header */
|
|
|
|
m_hdr->config_length = GV_GET64(be);
|
|
|
|
bcopy(d_hdr + 16, m_hdr->label.sysname, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(d_hdr + 48, m_hdr->label.name, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
m_hdr->label.date_of_birth.tv_sec = GV_GET64(be);
|
|
|
|
m_hdr->label.date_of_birth.tv_usec = GV_GET64(be);
|
|
|
|
m_hdr->label.last_update.tv_sec = GV_GET64(be);
|
|
|
|
m_hdr->label.last_update.tv_usec = GV_GET64(be);
|
|
|
|
m_hdr->label.drive_size = GV_GET64(be);
|
|
|
|
} else if (gv_legacy_header_type(d_hdr, be) == GV_LEGACY_POWERPC) {
|
2008-10-26 17:20:37 +00:00
|
|
|
G_VINUM_DEBUG(1, "detected legacy PowerPC header");
|
2008-10-01 14:50:36 +00:00
|
|
|
m_hdr->magic = GV_MAGIC;
|
|
|
|
/* legacy 32-bit big endian on-disk header */
|
|
|
|
m_hdr->config_length = GV_GET32(be);
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.sysname, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.name, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
m_hdr->label.date_of_birth.tv_sec = GV_GET32(be);
|
|
|
|
m_hdr->label.date_of_birth.tv_usec = GV_GET32(be);
|
|
|
|
m_hdr->label.last_update.tv_sec = GV_GET32(be);
|
|
|
|
m_hdr->label.last_update.tv_usec = GV_GET32(be);
|
|
|
|
m_hdr->label.drive_size = GV_GET64(be);
|
|
|
|
} else if (gv_legacy_header_type(d_hdr, be) == GV_LEGACY_I386) {
|
2008-10-26 17:20:37 +00:00
|
|
|
G_VINUM_DEBUG(1, "detected legacy i386 header");
|
2008-10-01 14:50:36 +00:00
|
|
|
m_hdr->magic = GV_MAGIC;
|
|
|
|
/* legacy i386 on-disk header */
|
|
|
|
m_hdr->config_length = GV_GET32(le);
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.sysname, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(d_hdr + off, m_hdr->label.name, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
m_hdr->label.date_of_birth.tv_sec = GV_GET32(le);
|
|
|
|
m_hdr->label.date_of_birth.tv_usec = GV_GET32(le);
|
|
|
|
m_hdr->label.last_update.tv_sec = GV_GET32(le);
|
|
|
|
m_hdr->label.last_update.tv_usec = GV_GET32(le);
|
|
|
|
m_hdr->label.drive_size = GV_GET64(le);
|
|
|
|
} else {
|
2008-10-26 17:20:37 +00:00
|
|
|
G_VINUM_DEBUG(1, "detected legacy amd64 header");
|
2008-10-01 14:50:36 +00:00
|
|
|
m_hdr->magic = GV_MAGIC;
|
|
|
|
/* legacy amd64 on-disk header */
|
|
|
|
m_hdr->config_length = GV_GET64(le);
|
|
|
|
bcopy(d_hdr + 16, m_hdr->label.sysname, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(d_hdr + 48, m_hdr->label.name, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
m_hdr->label.date_of_birth.tv_sec = GV_GET64(le);
|
|
|
|
m_hdr->label.date_of_birth.tv_usec = GV_GET64(le);
|
|
|
|
m_hdr->label.last_update.tv_sec = GV_GET64(le);
|
|
|
|
m_hdr->label.last_update.tv_usec = GV_GET64(le);
|
|
|
|
m_hdr->label.drive_size = GV_GET64(le);
|
|
|
|
}
|
|
|
|
|
|
|
|
g_free(d_hdr);
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Write out the gvinum header. */
|
|
|
|
int
|
|
|
|
gv_write_header(struct g_consumer *cp, struct gv_hdr *m_hdr)
|
|
|
|
{
|
|
|
|
uint8_t d_hdr[GV_HDR_LEN];
|
|
|
|
int off, ret;
|
|
|
|
|
|
|
|
#define GV_SET64BE(field) \
|
|
|
|
do { \
|
|
|
|
*((uint64_t *)&d_hdr[off]) = htobe64(field); \
|
|
|
|
off += 8; \
|
|
|
|
} while (0)
|
|
|
|
|
|
|
|
KASSERT(m_hdr != NULL, ("gv_write_header: null m_hdr"));
|
|
|
|
|
|
|
|
off = 0;
|
|
|
|
memset(d_hdr, 0, GV_HDR_LEN);
|
|
|
|
GV_SET64BE(m_hdr->magic);
|
|
|
|
GV_SET64BE(m_hdr->config_length);
|
|
|
|
off = 16;
|
|
|
|
bcopy(m_hdr->label.sysname, d_hdr + off, GV_HOSTNAME_LEN);
|
|
|
|
off += GV_HOSTNAME_LEN;
|
|
|
|
bcopy(m_hdr->label.name, d_hdr + off, GV_MAXDRIVENAME);
|
|
|
|
off += GV_MAXDRIVENAME;
|
|
|
|
GV_SET64BE(m_hdr->label.date_of_birth.tv_sec);
|
|
|
|
GV_SET64BE(m_hdr->label.date_of_birth.tv_usec);
|
|
|
|
GV_SET64BE(m_hdr->label.last_update.tv_sec);
|
|
|
|
GV_SET64BE(m_hdr->label.last_update.tv_usec);
|
|
|
|
GV_SET64BE(m_hdr->label.drive_size);
|
|
|
|
|
|
|
|
ret = g_write_data(cp, GV_HDR_OFFSET, d_hdr, GV_HDR_LEN);
|
|
|
|
return (ret);
|
|
|
|
}
|
2004-06-12 21:16:10 +00:00
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
/* Save the vinum configuration back to each involved disk. */
|
2004-08-27 21:32:18 +00:00
|
|
|
void
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
gv_save_config(struct gv_softc *sc)
|
2004-06-12 21:16:10 +00:00
|
|
|
{
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
struct g_consumer *cp;
|
2004-06-12 21:16:10 +00:00
|
|
|
struct gv_drive *d;
|
|
|
|
struct gv_hdr *vhdr, *hdr;
|
|
|
|
struct sbuf *sb;
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
struct timeval last_update;
|
2004-06-12 21:16:10 +00:00
|
|
|
int error;
|
|
|
|
|
|
|
|
KASSERT(sc != NULL, ("gv_save_config: null sc"));
|
|
|
|
|
|
|
|
vhdr = g_malloc(GV_HDR_LEN, M_WAITOK | M_ZERO);
|
|
|
|
vhdr->magic = GV_MAGIC;
|
|
|
|
vhdr->config_length = GV_CFG_LEN;
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
microtime(&last_update);
|
2004-06-12 21:16:10 +00:00
|
|
|
|
|
|
|
sb = sbuf_new(NULL, NULL, GV_CFG_LEN, SBUF_FIXEDLEN);
|
|
|
|
gv_format_config(sc, sb, 1, NULL);
|
|
|
|
sbuf_finish(sb);
|
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
LIST_FOREACH(d, &sc->drives, drive) {
|
|
|
|
/*
|
|
|
|
* We can't save the config on a drive that isn't up, but
|
|
|
|
* drives that were just created aren't officially up yet, so
|
|
|
|
* we check a special flag.
|
|
|
|
*/
|
|
|
|
if (d->state != GV_DRIVE_UP)
|
2004-06-12 21:16:10 +00:00
|
|
|
continue;
|
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
cp = d->consumer;
|
|
|
|
if (cp == NULL) {
|
|
|
|
G_VINUM_DEBUG(0, "drive '%s' has no consumer!",
|
|
|
|
d->name);
|
2004-09-13 21:01:36 +00:00
|
|
|
continue;
|
2004-06-12 21:16:10 +00:00
|
|
|
}
|
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
hdr = d->hdr;
|
|
|
|
if (hdr == NULL) {
|
|
|
|
G_VINUM_DEBUG(0, "drive '%s' has no header",
|
|
|
|
d->name);
|
2005-11-24 15:11:41 +00:00
|
|
|
g_free(vhdr);
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
continue;
|
2005-11-24 15:11:41 +00:00
|
|
|
}
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
bcopy(&last_update, &hdr->label.last_update,
|
|
|
|
sizeof(struct timeval));
|
|
|
|
bcopy(&hdr->label, &vhdr->label, sizeof(struct gv_label));
|
2005-08-26 14:40:32 +00:00
|
|
|
g_topology_lock();
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
error = g_access(cp, 0, 1, 0);
|
|
|
|
if (error) {
|
|
|
|
G_VINUM_DEBUG(0, "g_access failed on "
|
|
|
|
"drive %s, errno %d", d->name, error);
|
|
|
|
g_topology_unlock();
|
|
|
|
continue;
|
2004-06-12 21:16:10 +00:00
|
|
|
}
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
g_topology_unlock();
|
2004-06-12 21:16:10 +00:00
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
error = gv_write_header(cp, vhdr);
|
|
|
|
if (error) {
|
|
|
|
G_VINUM_DEBUG(0, "writing vhdr failed on drive %s, "
|
|
|
|
"errno %d", d->name, error);
|
|
|
|
g_topology_lock();
|
|
|
|
g_access(cp, 0, -1, 0);
|
|
|
|
g_topology_unlock();
|
|
|
|
continue;
|
2004-06-12 21:16:10 +00:00
|
|
|
}
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
/* First config copy. */
|
|
|
|
error = g_write_data(cp, GV_CFG_OFFSET, sbuf_data(sb),
|
|
|
|
GV_CFG_LEN);
|
|
|
|
if (error) {
|
|
|
|
G_VINUM_DEBUG(0, "writing first config copy failed on "
|
|
|
|
"drive %s, errno %d", d->name, error);
|
|
|
|
g_topology_lock();
|
|
|
|
g_access(cp, 0, -1, 0);
|
|
|
|
g_topology_unlock();
|
2004-06-12 21:16:10 +00:00
|
|
|
continue;
|
2005-05-17 16:38:30 +00:00
|
|
|
}
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
/* Second config copy. */
|
|
|
|
error = g_write_data(cp, GV_CFG_OFFSET + GV_CFG_LEN,
|
|
|
|
sbuf_data(sb), GV_CFG_LEN);
|
|
|
|
if (error)
|
|
|
|
G_VINUM_DEBUG(0, "writing second config copy failed on "
|
|
|
|
"drive %s, errno %d", d->name, error);
|
2005-05-17 16:38:30 +00:00
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
g_topology_lock();
|
|
|
|
g_access(cp, 0, -1, 0);
|
|
|
|
g_topology_unlock();
|
2005-05-17 16:38:30 +00:00
|
|
|
}
|
|
|
|
|
Import the gvinum work that have been done during and after Summer of Code 2007.
The work have been under testing and fixing since then, and it is mature enough
to be put into HEAD for further testing.
A lot have changed in this time, and here are the most important:
- Gvinum now uses one single workerthread instead of one thread for each
volume and each plex. The reason for this is that the previous scheme was
very complex, and was the cause of many of the bugs discovered in gvinum.
Instead, gvinum now uses one worker thread with an event queue, quite
similar to what used in gmirror.
- The rebuild/grow/initialize/parity check routines no longer runs in
separate threads, but are run as regular I/O requests with special flags.
This made it easier to support mounted growing and parity rebuild.
- Support for growing striped and raid5-plexes, meaning that one can extend the
volumes for these plex types in addition to the concat type. Also works while
the volume is mounted.
- Implementation of many of the missing commands from the old vinum:
attach/detach, start (was partially implemented), stop (was partially
implemented), concat, mirror, stripe, raid5 (shortcuts for creating volumes
with one plex of these organizations).
- The parity check and rebuild no longer goes between userland/kernel, meaning
that the gvinum command will not stay and wait forever for the rebuild to
finish. You can instead watch the status with the list command.
- Many problems with gvinum have been reported since 5.x, and some has been hard
to fix due to the complicated architecture. Hopefully, it should be more
stable and better handle edge cases that previously made gvinum crash.
- Failed drives no longer disappears entirely, but now leave behind a dummy
drive that makes sure the original state is not forgotten in case the system
is rebooted between drive failures/swaps.
- Update manpage to reflect new commands and extend it with some examples.
Sponsored by: Google Summer of Code 2007
Mentored by: le
Tested by: Rick C. Petty <rick-freebsd2008 -at- kiwi-computer.com>
2009-03-28 17:20:08 +00:00
|
|
|
sbuf_delete(sb);
|
|
|
|
g_free(vhdr);
|
2004-06-12 21:16:10 +00:00
|
|
|
}
|