]> sourceware.org Git - lvm2.git/log
lvm2.git
14 years agoAdd udev flags support in libdevmapper and provide 'dmsetup udevflags' command to...
Peter Rajnoha [Thu, 22 Oct 2009 12:55:47 +0000 (12:55 +0000)]
Add udev flags support in libdevmapper and provide 'dmsetup udevflags' command to decode them.

14 years ago.
Alasdair Kergon [Thu, 22 Oct 2009 11:25:46 +0000 (11:25 +0000)]
.

14 years agoOnly announce mirror monitoring to syslog after initialisation succeeds.
Petr Rockai [Thu, 22 Oct 2009 10:40:41 +0000 (10:40 +0000)]
Only announce mirror monitoring to syslog after initialisation succeeds.

14 years agoFix the "const discarded" warnings introduced in the unknown segment patch.
Petr Rockai [Thu, 22 Oct 2009 10:38:07 +0000 (10:38 +0000)]
Fix the "const discarded" warnings introduced in the unknown segment patch.

14 years agoClarify message:
Jonathan Earl Brassow [Wed, 21 Oct 2009 22:19:39 +0000 (22:19 +0000)]
Clarify message:
"  Mirror status: 1/2 images failed."

Half of an image?  Half of the images?  No, 1 of 2 images.

14 years agoHandle metadata with unknown segment types more gracefully.
Petr Rockai [Fri, 16 Oct 2009 17:41:49 +0000 (17:41 +0000)]
Handle metadata with unknown segment types more gracefully.

14 years agoI saw this in a bug report:
Jonathan Earl Brassow [Wed, 14 Oct 2009 14:55:44 +0000 (14:55 +0000)]
I saw this in a bug report:
[root@xxxx-01 ~]# lvconvert -m 1 --corelog VG/cmirror
  Unable to convert the log of inactive cluster mirror cmirror

I've tried to clean-up the message a little more, so the name
of the mirror stands out more while preserving the sense that
it's not a problem with the specific device, but the fact that
it is inactive that is causing the problem.

New msg:
  Unable to convert the log of an inactive cluster mirror, cmirror

14 years agoconfigure.in: don't set default file owner and group
Fabio M. Di Nitto [Wed, 14 Oct 2009 04:10:41 +0000 (04:10 +0000)]
configure.in: don't set default file owner and group

Per discussion on lvm-devel mailing list and part of debian patch set,
don't set defaults for owner and group, since nobody seems to use them, and
still allow override.

14 years agoAttempt to build dmeventd.static.
Alasdair Kergon [Tue, 13 Oct 2009 02:35:26 +0000 (02:35 +0000)]
Attempt to build dmeventd.static.

14 years agoMore build cleanups (linker parameter ordering).
Alasdair Kergon [Tue, 13 Oct 2009 01:31:10 +0000 (01:31 +0000)]
More build cleanups (linker parameter ordering).

14 years agoDisable realtime support by default.
Fabio M. Di Nitto [Mon, 12 Oct 2009 16:59:20 +0000 (16:59 +0000)]
Disable realtime support by default.

This code is used only in a under development feature and it's not used
anywhere in the tree.

Allow to build it conditionally.

14 years agoMake clvmd return 0 on success rather than 1.
Christine Caulfield [Mon, 12 Oct 2009 08:33:30 +0000 (08:33 +0000)]
Make clvmd return 0 on success rather than 1.

14 years agoMerge Debian patch 05 debian: keep libdm-abi consistent.
Fabio M. Di Nitto [Mon, 12 Oct 2009 04:06:42 +0000 (04:06 +0000)]
Merge Debian patch 05 debian: keep libdm-abi consistent.

This appears to be the only user visible feature that can change libdm ABI
at build time.

Thanks to Bastian Blank for the patch.

14 years agoExtra libs must be appended, not defined first.
Alasdair Kergon [Fri, 9 Oct 2009 16:04:54 +0000 (16:04 +0000)]
Extra libs must be appended, not defined first.

14 years agolinking parms missing from o->so conversion.
Alasdair Kergon [Fri, 9 Oct 2009 15:55:31 +0000 (15:55 +0000)]
linking parms missing from o->so conversion.

14 years agoCleanup comment and some whitespace.
Dave Wysochanski [Tue, 6 Oct 2009 16:00:38 +0000 (16:00 +0000)]
Cleanup comment and some whitespace.

14 years agoAdd --pvmetadatacopies as a synonym for --metadatacopies in various commands.
Dave Wysochanski [Mon, 5 Oct 2009 20:55:56 +0000 (20:55 +0000)]
Add --pvmetadatacopies as a synonym for --metadatacopies in various commands.

Going forward, we would like to allow users to specify the total
number of metadatacopies in a VG rather than on a per-PV basis.  In
order to facilitate that, introduce --pvmetadatacopes to replace
--metadatacopies everywhere.  We still allow --metadatacopies for
pv commands, but require --pvmetadatacopies for vg commands.
Eventually we will introduce --vgmetadatacopies.  Once we do that,
we should either deprecate --metadatacopies or make it a synonym based
on the command (pvmetadatacopies for pv commands, and vgmetadatacopies
for vg commands).  The latter option would likely just require a simple
'strncpy' check against cmd->command->name to qualify the merge_synonym
call.

Update nightly tests to cover the pvmetadatacopies synonym.

Note that this patch is the result of an eariler review comment for
the implicit pvcreate patches.  Should apply cleanly on top of the
implicit pvcreate patches (I applied after patch 10/10 in that series).

NOTE: This patch will require --pvmetadatacopies for vgconvert as
--metadatacopies is no longer accepted.

14 years agoAdd --pvmetadatacopies as a synonym for --metadatacopies in various commands.
Dave Wysochanski [Mon, 5 Oct 2009 20:53:41 +0000 (20:53 +0000)]
Add --pvmetadatacopies as a synonym for --metadatacopies in various commands.

Going forward, we would like to allow users to specify the total
number of metadatacopies in a VG rather than on a per-PV basis.  In
order to facilitate that, introduce --pvmetadatacopes to replace
--metadatacopies everywhere.  We still allow --metadatacopies for
pv commands, but require --pvmetadatacopies for vg commands.
Eventually we will introduce --vgmetadatacopies.  Once we do that,
we should either deprecate --metadatacopies or make it a synonym based
on the command (pvmetadatacopies for pv commands, and vgmetadatacopies
for vg commands).  The latter option would likely just require a simple
'strncpy' check against cmd->command->name to qualify the merge_synonym
call.

Update nightly tests to cover the pvmetadatacopies synonym.

Note that this patch is the result of an eariler review comment for
the implicit pvcreate patches.  Should apply cleanly on top of the
implicit pvcreate patches (I applied after patch 10/10 in that series).

NOTE: This patch will require --pvmetadatacopies for vgconvert as
--metadatacopies is no longer accepted.

14 years agoUpdate help messages for vgextend / vgcreate to mention PHYSICAL DEVICE OPTIONS.
Dave Wysochanski [Mon, 5 Oct 2009 20:22:58 +0000 (20:22 +0000)]
Update help messages for vgextend / vgcreate to mention PHYSICAL DEVICE OPTIONS.

14 years agoAdd implicit tests for implicit pvcreate during vgcreate and vgextend.
Dave Wysochanski [Mon, 5 Oct 2009 20:04:40 +0000 (20:04 +0000)]
Add implicit tests for implicit pvcreate during vgcreate and vgextend.

Update tests and WHATS_NEW for implicit pvcreate support.

14 years agoUpdate vgcreate and vgextend man pages to clarify implicit pvcreate.
Dave Wysochanski [Mon, 5 Oct 2009 20:04:21 +0000 (20:04 +0000)]
Update vgcreate and vgextend man pages to clarify implicit pvcreate.

vgcreate and vgextend now implicitly call pvcreate on devices not
currently initialized for LVM use.  Previously these commands would
fail with an error, so clarify the new behavior in the man page.

14 years agoAdd implicit pvcreate support to vgcreate and vgextend.
Dave Wysochanski [Mon, 5 Oct 2009 20:04:08 +0000 (20:04 +0000)]
Add implicit pvcreate support to vgcreate and vgextend.

Adds implicit pvcreate support when calling vgcreate or vgextend with
device paths that are not yet PVs.  This changes the behavior of vgcreate
and vgextend from failing with an error message to implicitly pvcreating.

14 years agoMove pvcreate_validate_params into toollib to allow calling from mutiple tools.
Dave Wysochanski [Mon, 5 Oct 2009 20:03:54 +0000 (20:03 +0000)]
Move pvcreate_validate_params into toollib to allow calling from mutiple tools.

For implicit pvcreate support, we need to call this from vgcreate and vgextend,
so move it into toollib.

14 years agoRefactor vgcreate/vgextend validation of vgname/pvname(s).
Dave Wysochanski [Mon, 5 Oct 2009 20:03:37 +0000 (20:03 +0000)]
Refactor vgcreate/vgextend validation of vgname/pvname(s).

Decrement argc and increment argv in a consistent way to allow for later
code-sharing.  Should be no functional change.

14 years agoRefactor pvcreate - split pvcreate_validate_params into recovery/non-recovery.
Dave Wysochanski [Mon, 5 Oct 2009 20:03:25 +0000 (20:03 +0000)]
Refactor pvcreate - split pvcreate_validate_params into recovery/non-recovery.

Split pvcreate_validate_params into recovery and non-recovery parameters.
This is necessary so we can call the non-recovery validate function from
vgextend / vgcreate.  Note in the pvcreate tool case, we must call the
recovery validation function first (see treatment of pe_start and --zero),
and that we add a call to fill_default_pvcreate_params before the validation
functions.

14 years agoAllow calling fill_default_pvcreate_params from tools.
Dave Wysochanski [Mon, 5 Oct 2009 20:03:08 +0000 (20:03 +0000)]
Allow calling fill_default_pvcreate_params from tools.

We need defaults for pvcreate_params at a higher level - this will
allow us to use a common function from the tools to take defaults,
then fill in any non-defaults from the commandline.

Future patches will refactor vgcreate/vgextend to call this function
if one or more pvcreate parameters are given on the commandline.

14 years agoAdd pvcreate_params to vg_extend.
Dave Wysochanski [Mon, 5 Oct 2009 20:02:48 +0000 (20:02 +0000)]
Add pvcreate_params to vg_extend.

Another refactoring for implicit pvcreate support.  We need to get
the pvcreate parameters somehow to the vg_extend routine.  Options
seemed to be:
1. Attach the parameters to struct volume_group.  I personally
did not like this idea in most cases, though one could make an
agrument why it might be ok at least for some of the parameters
(e.g. metadatacopies).
2. Pass them in to the extend routine.  This second route seemed
to be the best approach given the constraints.

Future patches will parse the command line and fill in the actual
values for the pvcreate_single call.
Should be no functional change.

14 years agoAdd pvcreate_params to vg_extend_single_pv.
Dave Wysochanski [Mon, 5 Oct 2009 20:02:30 +0000 (20:02 +0000)]
Add pvcreate_params to vg_extend_single_pv.

Should be no functional change.  If this parameter is set to NULL, just fail
the extend if the device is not already a PV.  If non-NULL, try pvcreate_single
before failing.  Note that pvcreate_single() handles the log_error in case
of failure so we just return 0 if pvcreate_single() fails.

14 years agoRefactor vg_extend - add vg_extend_single_pv.
Dave Wysochanski [Mon, 5 Oct 2009 20:02:04 +0000 (20:02 +0000)]
Refactor vg_extend - add vg_extend_single_pv.

Simple refactor to setup future changes related to implicit pvcreates.
Should be no functional change.

14 years agoMore makefile cleaning up and fixing. (gentoo)
Alasdair Kergon [Mon, 5 Oct 2009 13:46:00 +0000 (13:46 +0000)]
More makefile cleaning up and fixing.  (gentoo)

14 years agoCorrect example.conf to indicate that lvm2 not lvm1 is the default format.
Alasdair Kergon [Mon, 5 Oct 2009 12:44:20 +0000 (12:44 +0000)]
Correct example.conf to indicate that lvm2 not lvm1 is the default format.
Remove an unused stray LVM1_SUPPORT ifdef.

14 years agoOnly include selinux libs in libdevmapper.pc when selinux build enabled.
Alasdair Kergon [Mon, 5 Oct 2009 12:11:30 +0000 (12:11 +0000)]
Only include selinux libs in libdevmapper.pc when selinux build enabled.

14 years agoAllow for a build directory separate from the source.
Alasdair Kergon [Fri, 2 Oct 2009 19:10:31 +0000 (19:10 +0000)]
Allow for a build directory separate from the source.

14 years agoUpdate distclean target for rename clogd to cmirrord. (2.02.52)
Zdenek Kabelac [Fri, 2 Oct 2009 11:41:54 +0000 (11:41 +0000)]
Update distclean target for rename clogd to cmirrord. (2.02.52)

14 years agoRemove daemons/clogd from distclean dirs (replaced with cmirrord)
Zdenek Kabelac [Fri, 2 Oct 2009 11:35:25 +0000 (11:35 +0000)]
Remove daemons/clogd from distclean dirs (replaced with cmirrord)

14 years agoRun distclean also for daemons/cmirrord
Zdenek Kabelac [Fri, 2 Oct 2009 09:47:11 +0000 (09:47 +0000)]
Run distclean also for daemons/cmirrord

14 years agoAdd LCK_CONVERT flag I missed on the last checkin.
Christine Caulfield [Thu, 1 Oct 2009 14:15:34 +0000 (14:15 +0000)]
Add LCK_CONVERT flag I missed on the last checkin.

14 years agoStop clvmd from automatically doing lock conversions. Now, if a lock
Christine Caulfield [Thu, 1 Oct 2009 14:14:17 +0000 (14:14 +0000)]
Stop clvmd from automatically doing lock conversions. Now, if a lock
is granted at one mode and an attempt to convert it wthout the LCK_CONVERT
flag set then it will return errno=EBUSY.

This fixes a pretty bad bug in which an LV could be activated exclusively on
one node and lvchange -ay on another would convert it to shared!

It might break some things in other areas, but I doubt it.

14 years agoAdd percent_range to copy_percent too.
Alasdair Kergon [Thu, 1 Oct 2009 01:04:27 +0000 (01:04 +0000)]
Add percent_range to copy_percent too.

14 years agoIntroduce percent_range_t and centralise snapshot full/mirror in-sync checks.
Alasdair Kergon [Thu, 1 Oct 2009 00:35:29 +0000 (00:35 +0000)]
Introduce percent_range_t and centralise snapshot full/mirror in-sync checks.

14 years agoMake poll_mirror_progress a function that can be replaced.
Alasdair Kergon [Wed, 30 Sep 2009 18:15:06 +0000 (18:15 +0000)]
Make poll_mirror_progress a function that can be replaced.

14 years agoFactor out poll_mirror_progress and introduce progress_t.
Alasdair Kergon [Wed, 30 Sep 2009 17:43:51 +0000 (17:43 +0000)]
Factor out poll_mirror_progress and introduce progress_t.

14 years agoUpdate nightly tests to deal with unit changes.
Dave Wysochanski [Wed, 30 Sep 2009 16:13:53 +0000 (16:13 +0000)]
Update nightly tests to deal with unit changes.

Now uppercase letters imply Si units, so use lowercase everywhere.
We could stay with uppercase, but then we'd have to deal with rounding, etc.
Also, some output / error messages change slightly (instead of "GB" we're
now saying "GiB").
One test enhancement might be to add some new tests for the units changes
but for now let's just get the test back to passing.

14 years agoDistinguish between powers of 1000 and powers of 1024 in unit suffixes.
Alasdair Kergon [Wed, 30 Sep 2009 14:19:00 +0000 (14:19 +0000)]
Distinguish between powers of 1000 and powers of 1024 in unit suffixes.

14 years agoJust a cleanup from previous commit. We don't need pvname local var in _activate_lvs_...
Peter Rajnoha [Wed, 30 Sep 2009 12:05:25 +0000 (12:05 +0000)]
Just a cleanup from previous commit. We don't need pvname local var in _activate_lvs_in_vg anymore...

14 years agolook up pvmove by pvmove LV when PVMOVE flag is set
Alasdair Kergon [Tue, 29 Sep 2009 20:33:49 +0000 (20:33 +0000)]
look up pvmove by pvmove LV when PVMOVE flag is set

14 years agoDon't attempt to restart pvmoves when deactivating LVs in vgchange.
Alasdair Kergon [Tue, 29 Sep 2009 20:22:35 +0000 (20:22 +0000)]
Don't attempt to restart pvmoves when deactivating LVs in vgchange.
Restart lvconverts in vgchange by sharing lv_spawn_background_polling.

14 years agoGeneralise polldaemon code by changing mirror-specific variable names.
Alasdair Kergon [Tue, 29 Sep 2009 19:35:26 +0000 (19:35 +0000)]
Generalise polldaemon code by changing mirror-specific variable names.

14 years agoDon't attempt to deactivate an LV if any of its snapshots are in use.
Alasdair Kergon [Tue, 29 Sep 2009 18:50:28 +0000 (18:50 +0000)]
Don't attempt to deactivate an LV if any of its snapshots are in use.

14 years agoReturn fail if lv_deactivate did not removed device from kernel.
Milan Broz [Tue, 29 Sep 2009 15:17:54 +0000 (15:17 +0000)]
Return fail if lv_deactivate did not removed device from kernel.

lv_deactivate now returns always success, because tree deactivation
functions (see dm_tree_deactivate_children) always returns success.

Because code should return failure in lv_deactivate at least,
fix it by checking for device existence after real deactivation call.

(After discussion this was prefered solution to dm tree function rewrite
which affects snapshots and mirrors.)

14 years agoTrivial cleanup to lvcreate man page - use virtualsize in example.
Dave Wysochanski [Tue, 29 Sep 2009 15:11:06 +0000 (15:11 +0000)]
Trivial cleanup to lvcreate man page - use virtualsize in example.

virtualsize replaced virtualoriginsize but the example was never updated.

14 years agoProvide alternative implementation of obsolete siginterrupt().
Alasdair Kergon [Mon, 28 Sep 2009 21:23:02 +0000 (21:23 +0000)]
Provide alternative implementation of obsolete siginterrupt().

14 years agoConsolidate LV allocation into alloc_lv().
Alasdair Kergon [Mon, 28 Sep 2009 17:46:15 +0000 (17:46 +0000)]
Consolidate LV allocation into alloc_lv().

14 years agoTreat input units of both 's' and 'S' as 512-byte sectors. (2.02.49)
Alasdair Kergon [Mon, 28 Sep 2009 16:36:03 +0000 (16:36 +0000)]
Treat input units of both 's' and 'S' as 512-byte sectors.  (2.02.49)

's' and 'S' are special suffixes representing sectors and are always 512 bytes,
regardless of whether you're using powers of 1000 or 1024.

14 years agoAdd global/si_unit_consistency to enable cleaned-up use of units in output.
Alasdair Kergon [Mon, 28 Sep 2009 16:23:44 +0000 (16:23 +0000)]
Add global/si_unit_consistency to enable cleaned-up use of units in output.
Add configure --enable-units-compat to set si_unit_consistency off by default.

Use standard output units for 'PE Size' and 'Stripe size' in pv/lvdisplay.

14 years agopost-release
Alasdair Kergon [Sat, 26 Sep 2009 00:42:47 +0000 (00:42 +0000)]
post-release

14 years agoFix log fn prototype.
Alasdair Kergon [Sat, 26 Sep 2009 00:29:13 +0000 (00:29 +0000)]
Fix log fn prototype.

14 years agomissing dm_snprintf
Alasdair Kergon [Fri, 25 Sep 2009 19:06:05 +0000 (19:06 +0000)]
missing dm_snprintf

14 years agopre-release
Alasdair Kergon [Fri, 25 Sep 2009 18:30:27 +0000 (18:30 +0000)]
pre-release

14 years agopre-release
Alasdair Kergon [Fri, 25 Sep 2009 18:30:26 +0000 (18:30 +0000)]
pre-release

14 years agoensure dm_strdup succeeds
Alasdair Kergon [Fri, 25 Sep 2009 18:19:09 +0000 (18:19 +0000)]
ensure dm_strdup succeeds

14 years agoremove unused var & rename fn
Alasdair Kergon [Fri, 25 Sep 2009 18:13:17 +0000 (18:13 +0000)]
remove unused var & rename fn

14 years agoHandle any path supplied to dm_task_set_name() by looking up in /dev/mapper.
Alasdair Kergon [Fri, 25 Sep 2009 18:08:04 +0000 (18:08 +0000)]
Handle any path supplied to dm_task_set_name() by looking up in /dev/mapper.

14 years agoUse the same default umask for /dev dirs (DM_DEV_DIR_UMASK).
Peter Rajnoha [Fri, 25 Sep 2009 11:58:00 +0000 (11:58 +0000)]
Use the same default umask for /dev dirs (DM_DEV_DIR_UMASK).

14 years agoAdd more hints in 12-dm-permissions.rules. Add 'dmsetup splitname' and 'y|--yes'...
Peter Rajnoha [Wed, 23 Sep 2009 12:52:52 +0000 (12:52 +0000)]
Add more hints in 12-dm-permissions.rules. Add 'dmsetup splitname' and 'y|--yes' to dmsetup manpage.

14 years ago'_emit_areas_line' returns 1 for success and 0 for failure. This always
Jonathan Earl Brassow [Tue, 22 Sep 2009 16:26:59 +0000 (16:26 +0000)]
'_emit_areas_line' returns 1 for success and 0 for failure.  This always
confuses me, so I've added a comment at the top of the function to
remind me of this.

I also found that 'mirror_emit_segment_line' was returning 0 (return_0)
on failure /and/ success.  It now returns 1 for success and 0 for failure -
just like '_emit_areas_line' and its calling function, '_emit_segment_line'.

14 years agoMinor replace of of ${top_srcdir} with $(top_srcdir) for lcov target
Zdenek Kabelac [Tue, 22 Sep 2009 13:32:39 +0000 (13:32 +0000)]
Minor replace of of ${top_srcdir} with $(top_srcdir) for lcov target

14 years agoAdd some advisory comments for users in 12-dm-permissions.rules.
Peter Rajnoha [Tue, 22 Sep 2009 12:03:32 +0000 (12:03 +0000)]
Add some advisory comments for users in 12-dm-permissions.rules.

14 years agoFix dmeventd _temporary_log_fn parameters. (2.02.50)
Alasdair Kergon [Thu, 17 Sep 2009 10:37:23 +0000 (10:37 +0000)]
Fix dmeventd _temporary_log_fn parameters. (2.02.50)

14 years agoEnable dmeventd monitoring section of config file by default.
Alasdair Kergon [Wed, 16 Sep 2009 23:48:41 +0000 (23:48 +0000)]
Enable dmeventd monitoring section of config file by default.

14 years agodrop -ing suffix
Alasdair Kergon [Wed, 16 Sep 2009 23:40:19 +0000 (23:40 +0000)]
drop -ing suffix

14 years agoUpdate lvm2_monitoring script.
Alasdair Kergon [Wed, 16 Sep 2009 23:22:40 +0000 (23:22 +0000)]
Update lvm2_monitoring script.

14 years agoFix lvm2app test to run under test/api subdirectory only when configured.
Dave Wysochanski [Tue, 15 Sep 2009 19:59:04 +0000 (19:59 +0000)]
Fix lvm2app test to run under test/api subdirectory only when configured.

14 years agoAdd vg_is_resizeable() and cleanup references.
Dave Wysochanski [Tue, 15 Sep 2009 18:35:13 +0000 (18:35 +0000)]
Add vg_is_resizeable() and cleanup references.

Clean up VG_RESIZEABLE flag by creating vg_is_resizeable().
Update comment - we no longer have ALLOW_RESIZEABLE.
Also use vg_is_exported() in one place missed by earlier patch.
Should be no functional change.

14 years agoRemove test/api which should only be under 'make test'.
Alasdair Kergon [Tue, 15 Sep 2009 15:37:12 +0000 (15:37 +0000)]
Remove test/api which should only be under 'make test'.

14 years agopost-release
Alasdair Kergon [Tue, 15 Sep 2009 13:54:28 +0000 (13:54 +0000)]
post-release

14 years agopre-release cleanup
Alasdair Kergon [Tue, 15 Sep 2009 13:49:10 +0000 (13:49 +0000)]
pre-release cleanup

14 years agomissing (C) reminders
Alasdair Kergon [Tue, 15 Sep 2009 12:51:28 +0000 (12:51 +0000)]
missing (C) reminders

14 years agodm release cleanup
Alasdair Kergon [Tue, 15 Sep 2009 11:41:38 +0000 (11:41 +0000)]
dm release cleanup

14 years agopre-release
Alasdair Kergon [Tue, 15 Sep 2009 10:57:16 +0000 (10:57 +0000)]
pre-release

14 years agoFix process_each_vg / _process_one_vg when vg_read() returns FAILED_LOCKING.
Dave Wysochanski [Tue, 15 Sep 2009 01:38:59 +0000 (01:38 +0000)]
Fix process_each_vg / _process_one_vg when vg_read() returns FAILED_LOCKING.

Remove the checks for vg_read_error() in most of the tools callback
functions and instead make the check in _process_one_vg() more general.

In all but vgcfgbackup, we do not want to proceed if we get any error
from vg_read().  In vgcfgbackup's case, we may proceed if the backup
is to proceed with inconsistent VGs.  This is a special case though,
and we mark it with the READ_ALLOW_INCONSISTENT flag passed to
process_each_vg (and subsequently to _process_one_vg).

NOTE: More cleanup is needed in the vg_read_error() path cases.
This patch is a start.

14 years agoMore cmirror makefile fixes from Fabio.
Alasdair Kergon [Mon, 14 Sep 2009 22:57:46 +0000 (22:57 +0000)]
More cmirror makefile fixes from Fabio.

14 years agoFix build failure when enabling dmeventd and applib.
Dave Wysochanski [Mon, 14 Sep 2009 22:56:27 +0000 (22:56 +0000)]
Fix build failure when enabling dmeventd and applib.

This patch fixes a build with options similar to the following:
./configure --enable-debug --enable-applib --enable-dmeventd --enable-cmdlib

14 years agoAdd lots of missing stack debug messages to tools.
Alasdair Kergon [Mon, 14 Sep 2009 22:47:49 +0000 (22:47 +0000)]
Add lots of missing stack debug messages to tools.
Make readonly locking available as locking type 4.
Fix readonly locking to permit writeable global locks (for vgscan). (2.02.49)

14 years agoUse vg_is_exported(vg) macro everywhere.
Dave Wysochanski [Mon, 14 Sep 2009 19:44:15 +0000 (19:44 +0000)]
Use vg_is_exported(vg) macro everywhere.

This patch is all just cleanup and no other patch depends on it.
Replace explicit dereference and check with vg_is_exported().
Update a few copyrights and remove unnecessary whitespace.
Should be no functional change.

14 years agoAdd most relevant vg_attr fields as lvm2app 'get' functions.
Dave Wysochanski [Mon, 14 Sep 2009 19:43:11 +0000 (19:43 +0000)]
Add most relevant vg_attr fields as lvm2app 'get' functions.

Of the vgs field vg_attr, a few of the most likely to be used attributes
are clustered, exported, and partial.  This patch adds the following 3
functions:
uint64_t lvm_vg_is_clustered(const vg_t vg)
uint64_t lvm_vg_is_exported(const vg_t vg)
uint64_t lvm_vg_is_partial(const vg_t vg)

14 years agoUpdate lvm and vgs man pages to clarify 'partial' is a VG missing PVs.
Dave Wysochanski [Mon, 14 Sep 2009 19:42:13 +0000 (19:42 +0000)]
Update lvm and vgs man pages to clarify 'partial' is a VG missing PVs.

14 years agoAdd max_pv and max_lv vg 'get' lvm2app exports.
Dave Wysochanski [Mon, 14 Sep 2009 15:45:23 +0000 (15:45 +0000)]
Add max_pv and max_lv vg 'get' lvm2app exports.

14 years agoDo not run mdadm tests if MD module is not loaded.
Milan Broz [Sun, 13 Sep 2009 19:28:00 +0000 (19:28 +0000)]
Do not run mdadm tests if MD module is not loaded.

14 years agoMove dm_cookie_supported check into dm_udev_get_sync_supprt function.
Peter Rajnoha [Fri, 11 Sep 2009 16:11:25 +0000 (16:11 +0000)]
Move dm_cookie_supported check into dm_udev_get_sync_supprt function.

We don't have to call dm_cookie_supported with dm_udev_get_sync_support
this way. Also, it's necessary for the current code to work properly on
systems without cookie support (older kernels).

14 years agoAdd manpage entry for dmsetup's udevcomplete_all and udevcookies commands.
Peter Rajnoha [Fri, 11 Sep 2009 16:06:31 +0000 (16:06 +0000)]
Add manpage entry for dmsetup's udevcomplete_all and udevcookies commands.

14 years agoSeveral changes in udev rules:
Peter Rajnoha [Fri, 11 Sep 2009 16:05:20 +0000 (16:05 +0000)]
Several changes in udev rules:

- add DM_UDEV_RULES_VSN to provide a variable to be checked for in the other
  rules (e.g. to check that DM rules are actually installed, we can alternate
  functionality in the other rules based on this information, also we have
  versioning support for the rules)
- set proper sbin path for dmsetup and blkid, /sbin first, then /usr/sbin.
  This is necessary for anaconda to work properly.
- add 'last_rule' for cryptsetup's temporary devices (symlinks in /dev/mapper
  only)

14 years agoAdd one define that is necessary for older (experimental) libudev to work.
Peter Rajnoha [Fri, 11 Sep 2009 15:57:51 +0000 (15:57 +0000)]
Add one define that is necessary for older (experimental) libudev to work.

14 years agoCheck that udev is running and set internal state appropriately.
Peter Rajnoha [Fri, 11 Sep 2009 15:56:06 +0000 (15:56 +0000)]
Check that udev is running and set internal state appropriately.

14 years agoAdd libudev configuration check.
Peter Rajnoha [Fri, 11 Sep 2009 15:55:07 +0000 (15:55 +0000)]
Add libudev configuration check.

14 years agoAdd y|--yes option for dmsetup to provide a default 'YES' answer to questions.
Peter Rajnoha [Fri, 11 Sep 2009 15:53:57 +0000 (15:53 +0000)]
Add y|--yes option for dmsetup to provide a default 'YES' answer to questions.

14 years agoFix Makefile to take into account dmsetup's reconfiguration.
Peter Rajnoha [Fri, 11 Sep 2009 15:52:22 +0000 (15:52 +0000)]
Fix Makefile to take into account dmsetup's reconfiguration.

14 years agoUpdate lvm2app unit test vgtest - fix remove bug.
Dave Wysochanski [Fri, 4 Sep 2009 19:17:46 +0000 (19:17 +0000)]
Update lvm2app unit test vgtest - fix remove bug.

We now must commit to disk after lvm_vg_remove().

14 years agoFix path to test/api/vgtest in nightly testsuite.
Dave Wysochanski [Fri, 4 Sep 2009 14:26:16 +0000 (14:26 +0000)]
Fix path to test/api/vgtest in nightly testsuite.

14 years agoRemove 'test' interative api unit tests from 'TARGET' line.
Dave Wysochanski [Fri, 4 Sep 2009 13:49:02 +0000 (13:49 +0000)]
Remove 'test' interative api unit tests from 'TARGET' line.

The test/api directory TARGET line will be reserved for non-interactive
unit tests.  Building the interactive test can still be done with "make test"
from the test/api dir.

This page took 0.060372 seconds and 5 git commands to generate.