1
0
mirror of git://sourceware.org/git/lvm2.git synced 2024-12-22 17:35:59 +03:00
Commit Graph

11563 Commits

Author SHA1 Message Date
Zdenek Kabelac
95b5d24f43 configure: relocate configure.h
Commit 9403edbb93 move location of
configure.h and lvm-version.h.

Let's try even better place then /conf dir which should be left
for user configurable files.

Put these files right into include dir.
2015-09-11 21:46:35 +02:00
Zdenek Kabelac
19443035a6 configure: correct message 2015-09-11 21:46:34 +02:00
David Teigland
8b8103efef man lvmlockd: move and update text about vgremove
The requirement to stop VGs before vgremove applies
to both sanlock and dlm VGs.
2015-09-11 14:34:22 -05:00
David Teigland
6bc3d72a65 lvmlockd: prevent vgremove of dlm VG while lockspace is used
This applies the same rule/logic to dlm VGs that has always
existed for sanlock VGs.  Allowing a dlm VG to be removed
while its lockspace was still running on other hosts largely
worked, but there were difficult problems if another VG with
the same name was recreated.  Forcing the VG lockspace to
be stopped, gives both sanlock and dlm VGs the same behavior.
2015-09-11 14:12:13 -05:00
David Teigland
854a559a49 lvmlockd: remove shortcut for lockspace thread cleanup
This shortcut was added for an odd case that I do not
believe is relevant any more.  Having an alternate
path for lockspace thread cleanup is a complication
that could lead to problems.
2015-09-11 13:20:20 -05:00
David Teigland
18dfbbb150 lvmlockd: optimize dlm global lockspace creation
Set a flag indicating that the dlm global lockspace
exists to avoid going through the entire process of
trying to add it before finding it exists.
2015-09-11 13:20:01 -05:00
David Teigland
0a26c20b88 lockd: fix rescanning VG metadata
The code was expecting the wrong return value from
compare_config, which returns 0 when equal.

This is a problem for a lockd VG using multiple PVs
when the VG needs to be rescanned.
2015-09-11 13:10:48 -05:00
Zdenek Kabelac
0889cff5d5 spec: upgrade
Upgrade spec rules to be in touch with recent code.
Provide services and proper postin/preun/postun scriplets.
2015-09-10 17:28:47 +02:00
Zdenek Kabelac
9b8c876293 tests: remove only existing loops
If loop device meanwhile dissapered, skip 'losetup -d' call.
2015-09-10 17:28:47 +02:00
Zdenek Kabelac
e94ab01940 tests: fix check lv_not_exist
Use 'not' to set proper 'expected return status'
so we do not 'leak' debug logs printed for failing case.
2015-09-10 17:28:47 +02:00
Zdenek Kabelac
54c982081f tests: check cache stripe and raid behavior
Somehow raid tests landed in plain cache - separte them out
so they properly check for  have_raid.

Check we do not support strip option with cache-pool creation.
2015-09-10 17:28:47 +02:00
Zdenek Kabelac
a631fa20d0 cache: disallow stripes/size for cache pool
ATM allocation can't handle stripping and cache pool allocation.
It's not yet even clear what should be actually result.
Until resolved, disable this option (it's been coredumping
inside allocation anyway).
2015-09-10 17:28:15 +02:00
Zdenek Kabelac
5911fa1d91 cache: warn if caching causes troubles
Certain stacks of cached LVs may have unexpected consequences.
So add a warning function called when LV is cached to detect
such caces and WARN user about them - the best we could do ATM.
2015-09-10 17:27:30 +02:00
Zdenek Kabelac
e1edb5676e lib: when moving segtypes, move LV bits
When we insert layer we also move status flag-bits for certain LV types,
so internal volume_group structure remains consistent.
(Perhaps it's misuse of 'insert_layer' function and we should have
another similar function for this.)

Basically we aim to maintain the same state as after reading fresh
metadata out of volume group.

Currently we when i.e. cache  'raid' LV - this should transfer 'raidLV' flag
to  _corigin LV and cache is no longer a raid.

TODO: bits for stacked devices needs more exact rules.
2015-09-10 17:25:28 +02:00
David Teigland
3670f095c7 lvmlockd: check all variations of lvb values
The dlm will often lose the lvb content, so we need to
check quite a few possibilities for lvb values that
were not being checked before.

Refactoring was required to pass the entire lvb value
back to the core code instead of the single value.

The only functional change should be detecting new
lvb states where metadata is now invalidated where
it wasn't before.
2015-09-10 09:47:26 -05:00
David Teigland
f11d690967 lvmlockd: flag for internal actions
When an action is created by lvmlockd for itself,
there is no client to send the result to.  Add
the NO_CLIENT flag to the action to skip sending
the result to a client.
2015-09-10 09:47:26 -05:00
David Teigland
15ae237d2c lvmlockd: rename ADOPT_CLIENT_ID
to INTERNAL_CLIENT_ID since it will be
used for more than adopting locks.
2015-09-10 09:47:26 -05:00
David Teigland
36d16fed1f lockd: add start_init arg to lockd_start_vg
Add a new arg to lockd_start_vg() that indicates
it is being called for a new lockd VG, so that
lvmlockd knows the lockspace being started is new.
(Will be used by a following commit.)
2015-09-10 09:47:26 -05:00
David Teigland
30e489db5e vgchange: lock-stop only needs shared lock
lock-stop is one of the vgchange options that
doesn't change the VG so we can override the
default ex lock and use sh.
2015-09-10 09:47:26 -05:00
Peter Rajnoha
2296999cf6 dev-cache: ignore persistent cache if configuration changed
Commit f6473baffc introduced a new
cmd->initialized variable to keep info about which parts of the
cmd_context have been initialized.

A part of this patch was also a change in refresh_filters fn
which checks for cmd->initialized.filters variable and it does
the filter refresh *only* if the filter has already been initialized
before otherwise it's a NOOP (before, the refresh_filters also
initialized filters as a side effect in case it had not been
initialized before which was not quite correct).

However, the commit f6473baffc
did not handle the case in which configuration changes
either via --config argument or when configuration file changed
and its timestamp was higher than the timestamp of the persistent
cache file - the /etc/lvm/cache/.cache.

This patch fixes this issue and it causes the init_filters fn
in lvm_run_command fn to be called with proper value of
"load_persistent_cache" switch even if the configuration changes,
hence causing the persistent cache file to be ignored in this
case.
2015-09-10 16:13:39 +02:00
David Teigland
d323acdfec man lvmlockd: fix typo 2015-09-09 15:43:28 -05:00
David Teigland
81b0e9de7c man lvmlockd: mention device used for first vgcreate 2015-09-09 15:21:10 -05:00
Zdenek Kabelac
587fd6a0e4 tests: ensure dd flushed all data before exit 2015-09-08 15:41:34 +02:00
Zdenek Kabelac
6cb7f21e38 tests: better check for compiled version
Use install /lib files to check for expected version.
2015-09-08 15:41:34 +02:00
Zdenek Kabelac
8ff43c3705 makefiles: ensure no old .h are left
Ensure make clean  cleans any left-over file from their previous
location so they are not in conflict with new ones.

Also hide error message when .commands file is not present.
2015-09-08 15:41:33 +02:00
Peter Rajnoha
026db90621 WHATS_NEW: commit b77497cbd8 2015-09-08 15:37:23 +02:00
Peter Rajnoha
b77497cbd8 filters: make sure regex filter is evaluated before any filter that needs disk access
The regex filter (controlled by devices/filter lvm.conf setting) was
evaluated as the very last filter. However, this is not optimal when
it comes to restricting disk access - users define devices/filter
as well as devices/global_filter to avoid this.

The devices/global_filter is already positioned at the beginning of the
filter chain. We need to do the same for devices/filter.

Filter chains before this patch:

  A: when lvmetad is not used:
       persistent_filter -> sysfs_filter -> global_regex_filter ->
       type_filter -> usable->filter -> mpath_component_filter ->
       partition_filter -> md_component_filter -> fw_raid_filter ->
       regex_filter

  B: when lvmetad is used:

    B1: to update lvmetad:
      sysfs_filter -> global_regex_filter -> type_filter ->
      usable_filter -> mpath_component_filter -> partition_filter ->
      md_component_filter -> fw_raid_filter

    B2: to retrieve info from lvmetad:
      persistent_filter -> usable_filter -> regex_filter

From the chain list above we can see that particularly in case when
lvmetad is not used, the regex filter is the very last one that is
processed. If lvmetad is used, it doesn't matter much as there's
the global_regex_filter which is used instead when updating lvmetad
and when retrieving info from lvmetad, putting regex_filter in front
of usable_filter wouldn't change much since usabled_filter is not
reading disks directly.

This patch puts the regex filter to the front even in case lvmetad
is not used, hence reinstating the state as it was before commit
a7be3b12df (which moved the regex_filter
position in the chain). Still, the arguments for the commit
a7be3b12df still apply and they're
still satisfied since component filters (MD, mpath...) are evaluated
first just before updating lvmetad.

So with this patch, we end up with:

  A: when lvmetad is not used:
       persistent_filter -> sysfs_filter -> global_regex_filter ->
       regex_filter -> type_filter -> usable->filter ->
       mpath_component_filter -> partition_filter ->
       md_component_filter -> fw_raid_filter

  B: when lvmetad is used:

    B1: to update lvmetad:
      sysfs_filter -> global_regex_filter -> type_filter ->
      usable_filter -> mpath_component_filter -> partition_filter ->
      md_component_filter -> fw_raid_filter

    B2: to retrieve info from lvmetad:
      persistent_filter -> regex_filter -> usable_filter

This way, specifying the regex_filter in non-lvmetad case causes
the devices to be filtered based on regex first before processing
any other filters which can access disks (like md_component_filter).

This patch also streamlines the code for better readability.
2015-09-08 15:28:10 +02:00
Zdenek Kabelac
596ec5c74b tests: skip raid testing on 4.[1,2] fc24 kernels
Hopefull 4.3 will be fixed and test will be updated to let
raid test running again.

Meanwhile using md-raid may effectively kill kernel,
so leave at least other tests running.
2015-09-08 12:12:38 +02:00
Zdenek Kabelac
0ec64370b2 specs: install lvmlockctl only when built
Move install line into proper section.
2015-09-07 23:27:50 +02:00
Bryn M. Reeves
d7f45ebca5 libdm: clean up stats local variable use 2015-09-07 20:14:53 +01:00
Bryn M. Reeves
daa94eb792 libdm: fix display of nsec suffixes in histogram strings 2015-09-07 20:14:53 +01:00
Bryn M. Reeves
5f990473e4 libdm: clean up _build_histogram_arg()
Split up _build_histogram_arg() into separate functions to allocate
and fill the histogram arg string and remove nested local variable
declarations from the parent function.
2015-09-07 19:30:03 +01:00
Bryn M. Reeves
4bc7a86f3a libdm: only free the first histogram explicitly (Coverity)
Coverity flags a user-after-free in _stats_histograms_destroy():

>>>     Calling "dm_pool_free" frees pointer "mem->chunk" which has
>>>     already been freed.

This should not be possible since the histograms are destroyed in
reverse order of allocation:

 203         for (n = _nr_areas_region(region) - 1; n; n--)
 204                 if (region->counters[n].histogram)
 205                         dm_pool_free(mem, region->counters[n].histogram);

It appears that Coverity is unaware that pool->chunk is updated
during the call to dm_pool_free() and valgrind flags no errors in
this function when called with multiple allocated histograms.

Since there is no actual need to free the histograms individually
in this way simplify the code and just free the first allocated
object (which will also free all later allocated histograms in a
single call).
2015-09-07 17:53:56 +01:00
Zdenek Kabelac
ffbf12504d cleanup: assign seg_name once 2015-09-07 17:44:08 +02:00
Zdenek Kabelac
330cad1567 specs: package lvmlockctl man page 2015-09-07 17:44:08 +02:00
Zdenek Kabelac
fa4d2ec241 tests: update install rules
Enhnace 'make install' rules for test suite (used for rpm packaging).
Install lvmlockd conf files.
Compile runner with correct CXXFLAGS.
2015-09-07 17:44:08 +02:00
Zdenek Kabelac
acfc56957c makefiles: do not generate tags for everyone
Create tags file only on request (make tags).
Also clean file with 'make clean'.
2015-09-07 17:44:04 +02:00
Zdenek Kabelac
3ba431e79e makefiles: deps depends on header links
Put include/.symlinks_created as a prerequisite for dep calc.
Otherwise if these are not generated and user enters tests subdir and
runs 'make' he just gets endless loop of dep calculation.
2015-09-07 17:43:27 +02:00
Zdenek Kabelac
d62448cb45 configure: regenerate 2015-09-07 17:43:27 +02:00
Zdenek Kabelac
1999e368f1 configure: simpler version reading
Avoid calling 'cat' when 'read' could handle it.

Also drop DM_LIB_VERSION from AC_SUBST as it's not used anywhere
in .in file.
2015-09-07 17:43:27 +02:00
Zdenek Kabelac
fc4f0d3fce configure: support --disable-dependency-tracking
One-time build may enjoy little speedup of build when dependencies are not
calculated.
2015-09-07 17:43:11 +02:00
Zdenek Kabelac
9403edbb93 configure: relocate generated headers
Relocate generated configure.h and lvm-version.h outside
of compilable .c source tree.

The reason is behind - when compiling in builddir != srcdir
the generated file in lib/misc/configure.h was used for all compiled
source file except ones located in lib/misc dir - those would have used
configure.h file located in this dir - if there have existed one (i.e.
from some other build)

This problem was only visible, when srcdir == buildir was used before
trying to use  srcdri != builddir  (as configure.h appeared then in
srcdir).
2015-09-07 17:40:58 +02:00
Bryn M. Reeves
ab1b54c3e3 libdm: fix dm_stats leak in dm_stats_create()
The histogram changes adds a new error path to dm_stats_create().
Make sure that the dm_stats handle is properly destroyed if we fail
to create the histogram pool and check for failures setting the
program_id.
2015-09-07 12:08:34 +01:00
Bryn M. Reeves
0f5933ecc1 libdm: handle pool errors in dm_histogram_to_string() 2015-09-07 12:01:20 +01:00
Bryn M. Reeves
e75b4bc2df libdm: check dm_pool_begin_object() return value. 2015-09-07 11:52:54 +01:00
Bryn M. Reeves
36b09fd147 libdm: add missing error handling in _stats_parse_histogram()
Since we are growing an object in the histogram pool the return
value of dm_pool_grow_object() must be checked and error paths need
to abandon the object before returning.
2015-09-07 11:44:53 +01:00
Bryn M. Reeves
a26523330e dmsetup: fix bounds leak in _do_stats_create_regions()
If we fail to create the DM_DEVICE_TABLE dm_task destroy the bounds
object before returning.
2015-09-07 11:14:28 +01:00
Alasdair G Kergon
2a022e9e6e post-release 2015-09-05 23:58:40 +01:00
Alasdair G Kergon
fb12308416 style: Standardise some error paths. 2015-09-05 23:56:30 +01:00
Alasdair G Kergon
f5a3b05c7a pre-release 2015-09-05 23:46:22 +01:00