If first disk of IMSM RAID1 is failed but still present in the system,
the array is not auto-assembled. Auto-assemble uses raid disk slot from
metadata to index disks. As it's not set, the valid disk is seen as a
replacement disk and its metadata is ignored. The problem is not
observed for other RAID levels as they have more than 2 disks -
replacement disks are only stored under uneven indexes so third disk
metadata is used in such scenario.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Reviewed-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Try to use the full line length and avoid breaking up lines excessively.
Equally break up lines that are too long for no reason.
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
When rebuild is initiated by the UEFI driver it is possible that the new
disk will not contain a valid ppl header. Just write the initial ppl
and don't abort assembly.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Use the first map to get the correct disk when rebuilding and not the
failed disk from the second map.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Set resync_start to 0 when starting a rebuilding array to make the
kernel perform ppl recovery before the rebuild.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If array was initially assembled with kernel without PPL support -
initial header was never written to the drive.
If initial resync was completed and system is rebooted to kernel with
PPL support - mdadm prevents from assembling normal clean array
due to lack of valid PPL.
Write empty header when assemble normal clean array, so the
its assamble is no longer blocked.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If validate_ppl_imsm() detects an invalid ppl header it will be
overwritten with a valid, empty ppl header. But if we are assembling an
array after unclean shutdown this will cause the kernel to skip resync
after ppl recovery. We don't want that because if there was an invalid
ppl it's best to assume that the ppl recovery is not enough to make the
array consistent and a full resync should be performed. So when
overwriting the invalid ppl add one ppl_header_entry with a wrong
checksum. This will prevent the kernel from skipping resync after ppl
recovery.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If raid memeber is not in sync - it is skipped during
enablement of PPL. This is not correct, since the drive that
we are currently recovering to does not have ppl_size and ppl_sector
properly set in sysfs.
Remove this skipping, so all drives are updated during turning on the PPL.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If user has array with single ppl -
update the metadata to use multiple ppls.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
PPL area should be cleared before creation/force assemble.
If the drive was used in other RAID array, it might contains PPL from it.
There is a risk that mdadm recognizes those PPLs and
refuses to assemble the RAID due to PPL conflict with created
array.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Change validation algorithm to check validity of multiple ppls that
are stored in PPL area.
If read error occurs during - treat the all PPLs as invalid -
there is no guarantee that this one was not latest. If the header CRC is
incorrect - assume that there are no further PPLs in PPL area.
If whole PPL area was written at least once - there is a possibility that
old PPL (with lower generation number) will follow the recent one
(with higest generation number). Compare those generation numbers to check
which PPL is latest.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Add interpreting new rwh_policy bits. Set PPL size as 1MB.
If new array with ppl is created - use new implementation of ppl by
default.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Add support for super1 with multiple ppls. Extend ppl area size to 1MB.
Use 1MB as default during creation. Always start array as single ppl -
if kernel is capable of multiple ppls and there is enough space reserved -
it will switch the policy during first metadata update.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Change the behavior of assemble and create for consistency-policy=ppl
for external metadata arrays. If the kernel does not support ppl, don't
abort but print a warning and start the array without ppl
(consistency-policy=resync). No change for native md arrays because the
kernel will not allow starting the array if it finds an unsupported
feature bit in the superblock.
In sysfs_add_disk() check consistency_policy in the mdinfo structure
that represents the array, not the disk and read the current consistency
policy from sysfs in mdmon's manage_member(). This is necessary to make
sysfs_add_disk() honor the actual consistency policy and not what is in
the metadata. Also remove all the places where consistency_policy is set
for a disk's mdinfo - it is a property of the array, not the disk.
Signed-off-by: Artur Paszkiewicz <artur.paszkiewicz@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Update the manpage in "SIZE CHANGES" against the md commit.
Commit: e8a27f836f165c26f867ece7f31eb5c811692319
(md/bitmap: disable bitmap_resize for file-backed bitmaps.)
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Grow: Changing component size must be larger than current
chunk size against stripe raids, otherwise Grow_reshape()
would set s->size to '0'.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
GET_MISMATCH option doesn't exist for RAID arrays without redundancy
so sysfs_read fails if this information is requested. Set options
according to the device using information from /proc/mdstat.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Add disk controller domain for nvme and vmd devices to prevent moving
spares between different domains.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If array is stopped during reshape and assembled again straight away,
reshape process in a background might still be running. systemd doesn't
start a new service if one already exists. If there is a race, previous
process might terminate and new one is not created. Reshape doesn't
continue after assemble.
Tell systemd to restart the service rather than just start it. It will
assure previous service is stopped first. If it's not running, stopping
has no effect and only new process is started.
Signed-off-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Creating an array by opening a block-device with major number of 9
will transparently load the md module if needed.
Creating an array by opening
/sys/module/md_mod/parameters/new_array
and writing to it won't, it will just fail if md_mod isn't loaded.
So when opening that file fails with ENOENT, run "modprobe md_mod" and
try again.
This fixes a bug whereby if you have "CREATE names=yes" in mdadm.conf,
and the md modules isn't loaded, then creating or assembling an
array will not honor the "names=yes" configuration.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
This is a test case for testing --readonly and
--readwrite feature, it covers common metadata
versions and raid levels.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
The tests/check can be removed due to check()
in new 'test' has covered various checking.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
It's time to get rid of the tests/testdev due
to it has covered by testdev() in new 'test'.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
it would complain "too many arguments" when array[] has
few members, it's proper to check whether or not array[0]
is null.
Reported-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
--examine-bitmap: the bitmap_file_open() shouldn't omit the
regular file descriptor when the bitmap is external mode.
Such as: ./mdadm -X /mnt/3
This commit is partial revert of commit 0a6bff09d4
(mdadm/util: unify fstat checking blkdev into function)
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Summary:
In Incremental.c:count_active(), max_events is tracked to show to
which devices are up to date. If a device has events==max_events+1,
getinfo_super() is called to reload the superblock from this
device. getinfo_super1() blindly set journal_clean to 0, which is
wrong.
This patch fixes this by tracking max_journal_events for all the
disks. After scanning all disks, journal_clean is set if
max_journal_events >= max_events-1.
Signed-off-by: Song Liu <songliubraving@fb.com>
Reviewed-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Containers used to be handled as active arrays because GET_ARRAY_INFO
ioctl returns valid structure for them. As containers appear as inactive
in sysfs, the output for detail command has changed.
Stop relying on inactive state for containers. Make the output look the
same as in mdadm 4.0.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Spare migration doesn't work for external metadata. mdadm skips
a container with spare device because it is inactive. It used to work
because GET_ARRAY_INFO ioctl returned valid structure for a container
and mdadm treated such response as active container. Current
implementation checks it in sysfs where container is shown as inactive.
Adapt sysfs implementation to work the same way as ioctl.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
GET_MISMATCH option doesn't exist for containers so sysfs_read fails if
this information is requested. Set options according to the device using
information from /proc/mdstat.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
mdadm uses smaller exit codes like 0,1,2,3,4.
Using ERANGE is inconsistent and not helpful.
So change it to a more consistent number.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If no bitmap is present, then the test
if (__le32_to_cpu(bsb->nodes) > 1)
accesses uninitialised memory. So move that test inside
a test for a bitmap being present.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
${DEVLINKS}
is not valid udev syntax, and is passed through uninterpreted.
$env{DEVLINKS}
or
%e{DEVLINKS}
is correct.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Add "\n" to the end of error messages which don't already
have one. Also spell "opened" correctly.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
When RAID10 loses 2 disks and it is still operational, it cannot be
rebuilt. The rebuild process starts for the first disk and completes,
however completion is not recorded in metadata. There is an assumption
that rebuild completion corresponds to transition from degraded to
normal state. It's not the case for 2-disk RAID10 as it's still degraded
after rebuild to first disk completes.
Check if disk rebuild flag is set in the second map and clear it. So far it
has been checked only in the first map (where it was not set). The flag in
the second map has not been cleared but rebuild completion dropped second
map so the problem was not visible.
If rebuild completion is notified and array still has failed disks and is in
degraded state, check first if rebuild position is really unset (the same
check as for array in normal state). If so, mark migration as done but don't
change array state (it should remain degraded). Update failed disk number.
On rebuild start don't clear the rebuild flag in the destination map for all
the drives because failed state is lost for one of them. Just do a copy of
a map and clear the flag in the destination map for the disk that goes into
rebuild. Similarily preserve the rebuild flag in the map during disk removal.
If the disk is missing on array start and migration has been in progress,
don't just cancel it. Check first if maybe one of the disks was not under
rebuild (rebuild flag present both in source and destination map). If so,
rebuild was running despite of failed disk so there is no need to cancel
migration.
Signed-off-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
The check for inactive array is incorrect as it compares it against
active array. Introduce a new function md_is_array_active so the check
is consistent across the code.
As the output contains list of disks in the array include this
information in sysfs read.
Signed-off-by: Mariusz Tkaczyk <mariusz.tkaczyk@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
For cluster scenario, add device is different with native raid.
When a node issues adds a device, then the node will broadcast
a message with UUID to other nodes in the cluster. If receiving
node can find the device with the specific UUID, it must confirm
the device, otherwise reports it is missing.
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Currently, --add-journal can be only used to recreate broken journal
for arrays with journal since creation. As the kernel code getting
more mature, this constraint is no longer necessary.
This patch allows --add-journal to add journal to array without
journal.
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Currently if metadata requires more then 1M,
data offset will be rounded down to closest MB.
This is not correct, since less then required space is reserved.
Always round data offset up to multiple of 1M.
Signed-off-by: Pawel Baldysiak <pawel.baldysiak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If symlink (e.g. /dev/md/raid) is passed as a parameter to mdadm --wait,
it fails as it's not able to find a corresponding entry in /proc/mdstat
output. Get parameter file major:minor and look for block device name in
sysfs. This commit is partial revert of commit 9e04ac1c43
("mdadm/util: unify stat checking blkdev into function").
Signed-off-by: Tomasz Majchrzak <tomasz.majchrzak@intel.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
If test mode has set as '--dev=disk', then users can specify
the argument of "--disks" to test a bunch of physical devices.
For example: ./test --dev=disk --disks=/dev/sda{2..15}
could execute all test cases on physical devices.
Currently, the --dev=disk mode would confilct with testdev()
in current test cases, thus ignore testing testdev() if has
set --dev=disk mode.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
It supports to specify the argument of "--raidtype"
to run the different raid level cases. Details refer
to the do_help() usage.
For example: ./test --raidtype=raid1
could execute all the raid1 test cases under tests/.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Adding functions:
die()
uniform the abnormal situations that have to abort.
check_env()
do various basic checking before running test suite.
save_log()
collect array infos, include of dmesg, superblock,
bitmap and /proc/mdstat.
main()
the core function of this script.
Improve functions:
cleanup()
clear dmesg and remove the /var/tmp/mdtest* files.
mdadm()
clear superblock once creating or building arrays
every time, because it's always creating arrays
many times in a test case.
check()
just tidy up with die(), didn't change code meanings.
testdev()
add checking $1 must be a block device, add 'return 0'
in final because this function exists in last line of
test case, such as tests/05r6tor0.
do_test()
add checking abnormal dmesg and changing log management.
do_help()
just recommend a better way to print Usage.
parse_args()
revamp and improve.
Delete function:
fast_sync()
It's no longer used, so get rid of it.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
In case to make codes neat, this commit didn't change
any codes, just tidy up and use Tab as code format.
Signed-off-by: Zhilong Liu <zlliu@suse.com>
Signed-off-by: Jes Sorensen <jsorensen@fb.com>
Mariusz Tkaczyk
Jes Sorensen
Artur Paszkiewicz
Pawel Baldysiak
Zhilong Liu
Tomasz Majchrzak
Xiao Ni
NeilBrown
Song Liu
Guoqing Jiang