]> sourceware.org Git - lvm2.git/log
lvm2.git
5 years agolvmlockd: Fix arguments when built without sanlock
Marian Csontos [Tue, 18 Dec 2018 15:59:48 +0000 (16:59 +0100)]
lvmlockd: Fix arguments when built without sanlock

(cherry picked from commit 24254ca86671ae49431e67788c039603fe4b79fb)

5 years agoRevert "tests: Remove unsupported mirrored mirrorlog"
Marian Csontos [Tue, 18 Dec 2018 09:55:35 +0000 (10:55 +0100)]
Revert "tests: Remove unsupported mirrored mirrorlog"

Incorrect cherry pick from another branch.

This reverts commit 07fc4866f040a382c39577fb739e010734d83e42.

5 years agobuild: Remove badly placed `@`
Marian Csontos [Tue, 18 Dec 2018 08:42:34 +0000 (09:42 +0100)]
build: Remove badly placed `@`

5 years agobuild: Remove reference to undefined @BUILD_LVMETAD@
Marian Csontos [Tue, 18 Dec 2018 08:36:48 +0000 (09:36 +0100)]
build: Remove reference to undefined @BUILD_LVMETAD@

5 years agotests: Remove unsupported mirrored mirrorlog
Marian Csontos [Tue, 18 Dec 2018 07:48:25 +0000 (08:48 +0100)]
tests: Remove unsupported mirrored mirrorlog

5 years agolvcreate/lvconvert: optionally reenable mirrored mirror log for testing purposes...
Heinz Mauelshagen [Mon, 17 Dec 2018 18:24:37 +0000 (19:24 +0100)]
lvcreate/lvconvert: optionally reenable mirrored mirror log for testing purposes only

This is a followup patch to commit edb72cb70cb8abd0acb92f07937715be2816d04a
to support related lvm2 test suite tests.

A 'global/support_mirrored_mirror_log' bool configuration variable gets
introduced allowing the creation of, or conversion to mirrored 'mirror'
logs if set.  The capability to create these in turn allows the rest of
the tests to perform activation of such existing LVs and their conversions
to disk/core 'mirror' logs.

Display a disclaimer warning if enabled that this is not for regular use.

Add definition of the enabled config option to respective test scripts.

Related: rhbz1643562

5 years agounit test: use_lvmetad replaced by event_activation
David Teigland [Mon, 17 Dec 2018 16:04:44 +0000 (10:04 -0600)]
unit test: use_lvmetad replaced by event_activation

5 years agomakefiles: also fix build of unit test
Zdenek Kabelac [Mon, 17 Dec 2018 10:51:17 +0000 (11:51 +0100)]
makefiles: also fix build of unit test

5 years agomakefile: fixes build for older system
Zdenek Kabelac [Mon, 17 Dec 2018 10:41:38 +0000 (11:41 +0100)]
makefile: fixes build for older system

With older gcc - we need to resolve symbols linked with devmapper-event
that is now using -ldevmapper.

Also add forgotten systemd library needed for dbus notification.

5 years agomakefiles: correcting login of makefile
Zdenek Kabelac [Sat, 15 Dec 2018 00:55:11 +0000 (01:55 +0100)]
makefiles: correcting login of makefile

Fixing some ordering issue with inclusion of common make.tmpl.
Correcting dependency calculation
Simplifying inclusive makefile

5 years agomakefiles: dm-tools improve Makefile
Zdenek Kabelac [Mon, 17 Dec 2018 01:07:29 +0000 (02:07 +0100)]
makefiles: dm-tools improve Makefile

5 years agomakefiles: allow to set LIBS in Makefiles
Zdenek Kabelac [Sun, 16 Dec 2018 14:55:19 +0000 (15:55 +0100)]
makefiles: allow to set LIBS in Makefiles

Avoid doing hard set of LIBS var,
so if the LIBS is set before 'include make.tmpl' it's not lost.
This gives better control over order of linked libraries.

5 years agomakefiles: no longer used define
Zdenek Kabelac [Fri, 14 Dec 2018 23:20:15 +0000 (00:20 +0100)]
makefiles: no longer used define

Not used anymore -D_BUILDING_LVM

5 years agomakefiles: missing cleaning
Zdenek Kabelac [Mon, 17 Dec 2018 01:04:11 +0000 (02:04 +0100)]
makefiles: missing cleaning

5 years agomakefiles: quite install
Zdenek Kabelac [Sat, 15 Dec 2018 00:47:24 +0000 (01:47 +0100)]
makefiles: quite install

5 years agomakefiles: drop unneeded LIBS add
Zdenek Kabelac [Fri, 14 Dec 2018 22:27:35 +0000 (23:27 +0100)]
makefiles: drop unneeded LIBS add

5 years agomakefiles: sort
Zdenek Kabelac [Fri, 14 Dec 2018 20:50:09 +0000 (21:50 +0100)]
makefiles: sort

5 years agocmirror: link with libdm
Zdenek Kabelac [Fri, 14 Dec 2018 21:17:08 +0000 (22:17 +0100)]
cmirror: link with libdm

Since there is very little change there will be any new devel going
to happing with cmirror - avoid eating extra disk space and link
with already installed libdm which implements all use basic
function of dm list

5 years agolvm: drop usage of dl library
Zdenek Kabelac [Fri, 14 Dec 2018 20:51:11 +0000 (21:51 +0100)]
lvm: drop usage of dl library

Since lvm no longer supports any dlopen-able plugins
(which in practice was never really usable) drop linking
with -ldl.

5 years agodmeventd: do not link internal libraries to plugins
Zdenek Kabelac [Fri, 14 Dec 2018 15:58:56 +0000 (16:58 +0100)]
dmeventd: do not link internal libraries to plugins

Avoid linking internal device-mapper and link dynamically libdevmapper.
This considerably reduced size of installed binaries.

5 years agodmeventd: fix linking with libdevmapper
Zdenek Kabelac [Fri, 14 Dec 2018 15:28:26 +0000 (16:28 +0100)]
dmeventd: fix linking with libdevmapper

Since dmeventd is 'libdm' based project, it needs to link
libdm library instead of its internal version

An external users may provide plugins loadeable by dmeventd.
So external user of libdevmapper-event library has no other option
then to link with released libdevmapper library.

The complexity comes with lvm2 plugins.
The lvm2 plugin itself uses internal version of device_mapper,
but libdevmapper-event usage is libdm based - so there needs to be avoided
any breakage on compatibility of internal i.e. dm_task_run structures.

TODO: most likely dmeventd itself should be moved into libdm/dm-tools dir,
and only lvm2 plugins should be created as part of lvm project,
but those still need to link with libdevmapper.

5 years agodmfilemapd: avoid linking with DL_LIBS
Zdenek Kabelac [Fri, 14 Dec 2018 15:27:27 +0000 (16:27 +0100)]
dmfilemapd: avoid linking with DL_LIBS

Dmfilemapd is not using dlopen so doesn't need to be linked with
DL_LIBS.

5 years agolvconvert: ensure proper init of pv_list
Zdenek Kabelac [Fri, 14 Dec 2018 20:53:34 +0000 (21:53 +0100)]
lvconvert: ensure proper init of pv_list

5 years agolvmdbusd: Update table lookup for state lv_attr
Tony Asleson [Tue, 11 Dec 2018 17:24:51 +0000 (11:24 -0600)]
lvmdbusd: Update table lookup for state lv_attr

5 years agolvmdbusd: Update table lookup for health lv_attr
Tony Asleson [Tue, 11 Dec 2018 17:23:56 +0000 (11:23 -0600)]
lvmdbusd: Update table lookup for health lv_attr

5 years agolvmdbusd: Handle missing lv_attr table lookups
Tony Asleson [Tue, 11 Dec 2018 17:21:50 +0000 (11:21 -0600)]
lvmdbusd: Handle missing lv_attr table lookups

If we don't know the meaning we will return the key with default text
instead of raising an exception and taking the daemon out in the
process.

Resolves: rhbz1657950

5 years agolvmdbusd: Dump blackbox newest first
Tony Asleson [Tue, 11 Dec 2018 19:58:23 +0000 (13:58 -0600)]
lvmdbusd: Dump blackbox newest first

When we get bug reports we may not get the entire log, so lets
dump the fight recorder from newest to oldest as the one we
are interested in was likely to be the last command run.

5 years agolvmanip: uninitialized members in struct pv_list (#10)
Ming-Hung Tsai [Fri, 14 Dec 2018 09:51:51 +0000 (01:51 -0800)]
lvmanip:  uninitialized members in struct pv_list (#10)

Scenario: Given an existed LV `lvol0`, I want to create another LV
on the PVs used by `lvol0`.

I use `build_parallel_areas_from_lv()` to obtain the `pv_list` of each segments.
However, the returned `pv_list` is not properly initialized, which causes
segfault in subsequent operations.

5 years agodebug: drop extra tracing
Zdenek Kabelac [Sat, 1 Dec 2018 20:39:42 +0000 (21:39 +0100)]
debug: drop extra tracing

Stack tracing after log_error() is not needed.

5 years agocleanup: some local headers first
Zdenek Kabelac [Fri, 14 Dec 2018 12:10:25 +0000 (13:10 +0100)]
cleanup: some local headers first

5 years agomakefiles: some leftovers from lvmetad
Zdenek Kabelac [Sat, 1 Dec 2018 16:34:20 +0000 (17:34 +0100)]
makefiles: some leftovers from lvmetad

5 years agomakefiles: local headers first
Zdenek Kabelac [Fri, 14 Dec 2018 12:13:45 +0000 (13:13 +0100)]
makefiles: local headers first

5 years agomakefiles: drop unneeded include path
Zdenek Kabelac [Fri, 14 Dec 2018 12:12:58 +0000 (13:12 +0100)]
makefiles: drop unneeded include path

5 years agomakefiles: correct libdm dependency
Zdenek Kabelac [Fri, 14 Dec 2018 12:12:27 +0000 (13:12 +0100)]
makefiles: correct libdm dependency

Correcting dependency tracking for libdm internal files.

5 years agoheaders: use full path header instead of -I directive
Zdenek Kabelac [Mon, 3 Dec 2018 17:15:44 +0000 (18:15 +0100)]
headers: use full path header instead of -I directive

Replace Makefile -I  with full path to headers like we use now
everywhere else.

5 years agoheaders: use configure.h as 1st. header
Zdenek Kabelac [Mon, 3 Dec 2018 17:14:26 +0000 (18:14 +0100)]
headers: use configure.h as 1st. header

Ensure configure.h is always 1st. included header.
Maybe we could eventually introduce gcc -include option, but for now
this better uses dependency tracking.

Also move _REENTRANT and _GNU_SOURCE into configure.h so it
doesn't need to be present in various source files.
This ensures consistent compilation of headers like stdio.h since
it may produce different declaration.

5 years agolibdm: use libdm header
Zdenek Kabelac [Fri, 14 Dec 2018 12:11:48 +0000 (13:11 +0100)]
libdm: use libdm header

User libdm header file.

5 years agodmeventd: Fix libdevmapper-event linking
Marian Csontos [Wed, 12 Dec 2018 09:50:36 +0000 (10:50 +0100)]
dmeventd: Fix libdevmapper-event linking

5 years agoman: document 's' RAID attribute bit
Heinz Mauelshagen [Tue, 11 Dec 2018 15:54:17 +0000 (16:54 +0100)]
man: document 's' RAID attribute bit

5 years agoraid: fix (de)activation of RaidLVs with visible SubLVs
Heinz Mauelshagen [Wed, 31 Oct 2018 22:05:08 +0000 (23:05 +0100)]
raid: fix (de)activation of RaidLVs with visible SubLVs

There's a small window during creation of a new RaidLV when
rmeta SubLVs are made visible to wipe them in order to prevent
erroneous discovery of stale RAID metadata.  In case a crash
prevents the SubLVs from being committed hidden after such
wiping, the RaidLV can still be activated with the SubLVs visible.
During deactivation though, a deadlock occurs because the visible
SubLVs are deactivated before the RaidLV.

The patch adds _check_raid_sublvs to the raid validation in merge.c,
an activation check to activate.c (paranoid, because the merge.c check
will prevent activation in case of visible SubLVs) and shares the
existing wiping function _clear_lvs in raid_manip.c moved to lv_manip.c
and renamed to activate_and_wipe_lvlist to remove code duplication.
Whilst on it, introduce activate_and_wipe_lv to share with
(lvconvert|lvchange).c.

Resolves: rhbz1633167

5 years agoWHATS_NEW
Heinz Mauelshagen [Mon, 10 Dec 2018 13:48:13 +0000 (14:48 +0100)]
WHATS_NEW

5 years agolvcreate/lvconvert: prohibit creation of/conversion to mirrored mirror logs
Heinz Mauelshagen [Fri, 7 Dec 2018 18:56:49 +0000 (19:56 +0100)]
lvcreate/lvconvert: prohibit creation of/conversion to mirrored mirror logs

In RHEL7 we marked mirrored mirror logs as deprecated and
added a related message.  This patch prohibits creating new
'mirror' LVs with that log type or converting existing LVs
to have one.

Existing LVs with mirrored mirror log can be activated
and converted to disk/core logs.

Avoid double deprecation message when running lvconvert.

Resolves: rhbz1643562

5 years agolvmlockd: fix missing LV lock for lvconvert repair
David Teigland [Fri, 7 Dec 2018 19:11:31 +0000 (13:11 -0600)]
lvmlockd: fix missing LV lock for lvconvert repair

Add missing lvmlockd LV lock for lvconvert repair
on mirror and thin/cache pools.

5 years agoudev: 69-dm-lvm-metad.rules is still needed
Marian Csontos [Thu, 6 Dec 2018 09:50:42 +0000 (10:50 +0100)]
udev: 69-dm-lvm-metad.rules is still needed

5 years agobuild: Upse PYTHON_CONFIG env.variable when set
Marian Csontos [Fri, 2 Nov 2018 09:06:51 +0000 (10:06 +0100)]
build: Upse PYTHON_CONFIG env.variable when set

This adds up to Commit 6462e8dffc7ed53973ce796cec18fecc0fb89d77.

5 years agotests: lvm-on-md udev issues
David Teigland [Wed, 5 Dec 2018 18:14:17 +0000 (12:14 -0600)]
tests: lvm-on-md udev issues

5 years agoman lvmlockd: lvextend with gfs2
David Teigland [Wed, 5 Dec 2018 17:31:58 +0000 (11:31 -0600)]
man lvmlockd: lvextend with gfs2

5 years agoremove unused full filter
David Teigland [Tue, 4 Dec 2018 20:06:46 +0000 (14:06 -0600)]
remove unused full filter

it's the same as cmd->filter

5 years agoremove unused lvmetad filter
David Teigland [Tue, 4 Dec 2018 18:44:43 +0000 (12:44 -0600)]
remove unused lvmetad filter

5 years agopvscan: use correct dev filters
David Teigland [Mon, 3 Dec 2018 18:58:46 +0000 (12:58 -0600)]
pvscan: use correct dev filters

pvscan was still using lvmetad_filter which has been
null since lvmetad was removed.  Switch it to use the
full_filter.

5 years agodevs: use udev info to improve md component detection
David Teigland [Mon, 3 Dec 2018 17:22:45 +0000 (11:22 -0600)]
devs: use udev info to improve md component detection

Use udev info to supplement native md component detection.

5 years agogcc: ensure sector is initilized
Zdenek Kabelac [Fri, 30 Nov 2018 21:26:26 +0000 (22:26 +0100)]
gcc: ensure sector is initilized

Some older gcc errnously report the variable can be used uninitlized.
Quite warning by explicit initalization.

5 years agogcc: avoid shadowing use_aio
Zdenek Kabelac [Fri, 30 Nov 2018 20:56:16 +0000 (21:56 +0100)]
gcc: avoid shadowing use_aio

Function use_aio() is already declared, avoid its shadowing.
lvm-globals.h:59: warning: shadowed declaration is here

5 years agogcc: avoid shadowing activate_lv
Zdenek Kabelac [Fri, 30 Nov 2018 20:50:41 +0000 (21:50 +0100)]
gcc: avoid shadowing activate_lv

Function activate_lv() is already declared, avoid its shadowing.
activate.h:133: warning: shadowed declaration is here

5 years agogcc: avoid shadowing index
Zdenek Kabelac [Fri, 30 Nov 2018 20:46:14 +0000 (21:46 +0100)]
gcc: avoid shadowing index

Some older headers were declaring 'index' so avoid its usage.
/usr/include/string.h:489: warning: shadowed declaration is here

5 years agomakefiles: ensure test dir can run unit-test
Zdenek Kabelac [Fri, 30 Nov 2018 23:44:22 +0000 (00:44 +0100)]
makefiles: ensure test dir can run unit-test

5 years agomakefiles: avoid clustering out
Zdenek Kabelac [Fri, 30 Nov 2018 20:59:08 +0000 (21:59 +0100)]
makefiles: avoid clustering out

When ctags package is not installed, avoid cluttering output
from failing 'which' command.

5 years agodevice_mapper: move internal header to front
Zdenek Kabelac [Fri, 30 Nov 2018 23:43:53 +0000 (00:43 +0100)]
device_mapper: move internal header to front

5 years agolibdm: optimize dm_pool_strndup
Zdenek Kabelac [Fri, 30 Nov 2018 23:35:41 +0000 (00:35 +0100)]
libdm: optimize dm_pool_strndup

5 years agodevice_mapper: optimize dm_pool_strndup
Zdenek Kabelac [Fri, 30 Nov 2018 23:34:29 +0000 (00:34 +0100)]
device_mapper: optimize dm_pool_strndup

5 years agodevice_mapper: fix incorrect dm_strncpy usage
Zdenek Kabelac [Fri, 30 Nov 2018 23:41:24 +0000 (00:41 +0100)]
device_mapper: fix incorrect dm_strncpy usage

Patch 668c9d0762fc6736a1e9372785544eda6a939806 introduced regression,
since the code here would actually always return failing result.
Replace it with more simple call to strndup().

5 years agorpm: install lvm2-pvscan again
Zdenek Kabelac [Fri, 30 Nov 2018 12:09:09 +0000 (13:09 +0100)]
rpm: install lvm2-pvscan again

Let's repeat history once more 13d5c78a8d5f09b91e26cd49fffa05313d2f6c82.

5 years agotests: extend sleep
Zdenek Kabelac [Fri, 30 Nov 2018 11:55:05 +0000 (12:55 +0100)]
tests: extend sleep

5 years agomakefiles: clean unit-test
Zdenek Kabelac [Fri, 30 Nov 2018 11:55:22 +0000 (12:55 +0100)]
makefiles: clean unit-test

5 years agomakefiles: ignore missing files
Zdenek Kabelac [Fri, 30 Nov 2018 11:54:34 +0000 (12:54 +0100)]
makefiles: ignore missing files

5 years agoconfigure: update
Zdenek Kabelac [Fri, 30 Nov 2018 12:02:42 +0000 (13:02 +0100)]
configure: update

5 years agotests: drop use_lvmetad from unit test
Zdenek Kabelac [Thu, 29 Nov 2018 21:52:46 +0000 (22:52 +0100)]
tests: drop use_lvmetad from unit test

5 years agotests: reduce memory footprint
Zdenek Kabelac [Tue, 27 Nov 2018 11:15:52 +0000 (12:15 +0100)]
tests: reduce memory footprint

5 years agotests: updates
Zdenek Kabelac [Fri, 23 Nov 2018 19:18:22 +0000 (20:18 +0100)]
tests: updates

5 years agotests: requires at least 2 iterations
Zdenek Kabelac [Fri, 23 Nov 2018 19:09:14 +0000 (20:09 +0100)]
tests: requires at least 2 iterations

5 years agotests: use select with dmsetup
Zdenek Kabelac [Fri, 23 Nov 2018 19:08:08 +0000 (20:08 +0100)]
tests: use select with dmsetup

Use 'dmsetup -S' to greatly simplify such loops.

5 years agotests: add mising udev_wait
Zdenek Kabelac [Tue, 20 Nov 2018 15:01:12 +0000 (16:01 +0100)]
tests: add mising udev_wait

mdadm does not handle udev waiting so it may exit earlier,
while devices are still 'running'.

5 years agomake: generate man update
Zdenek Kabelac [Thu, 29 Nov 2018 22:08:25 +0000 (23:08 +0100)]
make: generate man update

5 years agomake: generate config update
Zdenek Kabelac [Thu, 29 Nov 2018 22:08:05 +0000 (23:08 +0100)]
make: generate config update

5 years agomakefiles: add missing srcdir
Zdenek Kabelac [Thu, 29 Nov 2018 21:52:10 +0000 (22:52 +0100)]
makefiles: add missing srcdir

5 years agomakefiles: improving cleaning rules
Zdenek Kabelac [Thu, 29 Nov 2018 20:47:11 +0000 (21:47 +0100)]
makefiles: improving cleaning rules

5 years agomakefiles: avoid dependency calcs for base dir
Zdenek Kabelac [Thu, 29 Nov 2018 16:34:32 +0000 (17:34 +0100)]
makefiles: avoid dependency calcs for base dir

For some targets we do not want to generate dependencies.
Also add note about usage of such Makefile - it might be
possibly better to rename it to different filename to avoid
any confusion.

5 years agomakefiles: updates for less verbosity
Zdenek Kabelac [Thu, 29 Nov 2018 16:09:27 +0000 (17:09 +0100)]
makefiles: updates for less verbosity

5 years agomakefiles: improve lcov generator
Zdenek Kabelac [Fri, 23 Nov 2018 12:07:50 +0000 (13:07 +0100)]
makefiles: improve lcov generator

Simplify generation of lcov report.

5 years agobase: use calloc
Zdenek Kabelac [Tue, 27 Nov 2018 11:12:51 +0000 (12:12 +0100)]
base: use calloc

Make zalloc a wrapper over calloc

5 years agoscan: md metadata version 0.90 is at the end of disk
Peter Rajnoha [Thu, 29 Nov 2018 18:35:54 +0000 (12:35 -0600)]
scan: md metadata version 0.90 is at the end of disk

commit de28637
  scan: use full md filter when md 1.0 devices are present

missed the fact that md superblock version 0.90 also puts
metadata at the end of the device, so the full md filter
needs to be used when either 0.90 or 1.0 is present.

5 years agoconfig settings: fix version 3.0.0
David Teigland [Wed, 28 Nov 2018 18:16:50 +0000 (12:16 -0600)]
config settings: fix version 3.0.0

version 3.0.0 was changed in the end to 2.3.0,
but config settings had previously been encoded
with version 3.0.0.

5 years agolvmlockd: vgchange locktype with yes option
David Teigland [Tue, 27 Nov 2018 20:38:39 +0000 (14:38 -0600)]
lvmlockd: vgchange locktype with yes option

for auto response to yes/no prompt.

5 years agolvmlockctl: wait by default when stopping
David Teigland [Tue, 27 Nov 2018 19:41:33 +0000 (13:41 -0600)]
lvmlockctl: wait by default when stopping

lvmlockctl --stop-lockspaces was by default not waiting
for all the lockspaces to be gone.

5 years agoPlace the first PE at 1 MiB for all defaults
David Teigland [Tue, 13 Nov 2018 21:00:11 +0000 (15:00 -0600)]
Place the first PE at 1 MiB for all defaults

. When using default settings, this commit should change
  nothing.  The first PE continues to be placed at 1 MiB
  resulting in a metadata area size of 1020 KiB (for
  4K page sizes; slightly smaller for larger page sizes.)

. When default_data_alignment is disabled in lvm.conf,
  align pe_start at 1 MiB, based on a default metadata area
  size that adapts to the page size.  Previously, disabling
  this option would result in mda_size that was too small
  for common use, and produced a 64 KiB aligned pe_start.

. Customized pe_start and mda_size values continue to be
  set as before in lvm.conf and command line.

. Remove the configure option for setting default_data_alignment
  at build time.

. Improve alignment related option descriptions.

. Add section about alignment to pvcreate man page.

Previously, DEFAULT_PVMETADATASIZE was 255 sectors.
However, the fact that the config setting named
"default_data_alignment" has a default value of 1 (MiB)
meant that DEFAULT_PVMETADATASIZE was having no effect.

The metadata area size is the space between the start of
the metadata area (page size offset from the start of the
device) and the first PE (1 MiB by default due to
default_data_alignment 1.)  The result is a 1020 KiB metadata
area on machines with 4KiB page size (1024 KiB - 4 KiB),
and smaller on machines with larger page size.

If default_data_alignment was set to 0 (disabled), then
DEFAULT_PVMETADATASIZE 255 would take effect, and produce a
metadata area that was 188 KiB and pe_start of 192 KiB.
This was too small for common use.

This is fixed by making the default metadata area size a
computed value that matches the value produced by
default_data_alignment.

5 years agoblk_availability service drop lvmetad
David Teigland [Mon, 26 Nov 2018 20:51:32 +0000 (14:51 -0600)]
blk_availability service drop lvmetad

5 years agopvscan systemd service for event based activation
David Teigland [Mon, 26 Nov 2018 18:49:39 +0000 (12:49 -0600)]
pvscan systemd service for event based activation

The pvscan systemd service for autoactivation was
mistakenly dropped along with the lvmetad related
services.

The activation generator program now looks at the new
lvm.conf setting "event_activation" (default 1) to
switch between event activation and direct activation.

Previously, the old use_lvmetad setting was used to
switch between event and direct activation.

5 years agowritecache: set block_size using --cachesettings
David Teigland [Wed, 21 Nov 2018 21:16:23 +0000 (15:16 -0600)]
writecache: set block_size using --cachesettings

instead of a separate --writecacheblocksize option.
writecache block_size is not technically a setting,
but it can borrow the option as a special case.

5 years agoWHATS_NEW: sync io
David Teigland [Tue, 20 Nov 2018 15:20:28 +0000 (09:20 -0600)]
WHATS_NEW: sync io

5 years agobcache: sync io fixes
David Teigland [Fri, 16 Nov 2018 19:09:29 +0000 (13:09 -0600)]
bcache: sync io fixes

fix lseek error check
fix read/write error checks
handle zero return from read and write
don't return an error for short io
fix partial read/write loop

5 years agoio: use sync io if aio fails
David Teigland [Fri, 16 Nov 2018 18:21:20 +0000 (12:21 -0600)]
io: use sync io if aio fails

io_setup() for aio may fail if a system has reached the
aio request limit.  In this case, fall back to using
sync io.  Also, lvm use of aio can be disabled entirely
with config setting global/use_aio=0.

The system limit for aio requests can be seen from
  /proc/sys/fs/aio-max-nr

The current usage of aio requests can be seen from
  /proc/sys/fs/aio-nr

The system limit for aio requests can be increased by
setting fs.aio-max-nr using sysctl.

Also add last-byte limit to the sync io code.

5 years agotests: update required raid target
Zdenek Kabelac [Mon, 19 Nov 2018 17:06:39 +0000 (18:06 +0100)]
tests: update required raid target

For embeded reshaping operation require higher driver version.
(otherwise we get:

Converting vg/LV1 from raid6 (same as raid6_zr) is directly possible to the following layouts:
 raid6_nc
 raid6_nr
 raid6_la_6
 raid6_ls_6
 raid6_ra_6
 raid6_rs_6
 raid6_n_6

5 years agotests: missing copyright
Zdenek Kabelac [Mon, 19 Nov 2018 16:16:36 +0000 (17:16 +0100)]
tests: missing copyright

5 years agosanlock: update headers
Zdenek Kabelac [Mon, 19 Nov 2018 15:55:11 +0000 (16:55 +0100)]
sanlock: update headers

5 years agotests: generate slightly less volumes
Zdenek Kabelac [Mon, 19 Nov 2018 12:39:39 +0000 (13:39 +0100)]
tests: generate slightly less volumes

5 years agotests: speed-up testing full of lvm2 metadata
Zdenek Kabelac [Mon, 19 Nov 2018 12:15:05 +0000 (13:15 +0100)]
tests: speed-up testing full of lvm2 metadata

Generate faster full metadata condition.

FIXME: vgremove is extremaly slow with larger set of LVs.

5 years agotests: update parm for new kernel
Zdenek Kabelac [Sun, 18 Nov 2018 21:04:00 +0000 (22:04 +0100)]
tests: update parm for new kernel

5 years agotests: skip part of test
Zdenek Kabelac [Sun, 18 Nov 2018 21:01:27 +0000 (22:01 +0100)]
tests: skip part of test

On kernel 4.8 this test piece oopses machine (fc23).

5 years agotests: create whole path with mkdir
Zdenek Kabelac [Sun, 18 Nov 2018 20:42:27 +0000 (21:42 +0100)]
tests: create whole path with mkdir

Create also 'lvm' upper this if such one is missing in /run dir
and use '-p' with mkdir.

5 years agotests: skip portion of test for lvmpolld
Zdenek Kabelac [Sun, 18 Nov 2018 20:38:56 +0000 (21:38 +0100)]
tests: skip portion of test for lvmpolld

lvmpolld ATM is not desingned to preserve interval checking
in the same way the 'lvconvert' tool is doing - so the passed
'-i 40' is not respected and lvmpolld autonomously checks
state of conversion and updates lvm2 metadata and dm tables
when needed.

So skip portion of test that relayed on this and preserve this logic
only for command line invocation and forking of polling process
where the interval of checking is under full control.

This page took 0.078199 seconds and 5 git commands to generate.