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

7596 Commits

Author SHA1 Message Date
Petr Rockai
93108125ac test: Fix fallout from pv_read changes. 2013-06-05 12:37:37 +02:00
Petr Rockai
d17cbf112b pvresize: Do not use pv_read (get the PV from orphan VG). 2013-06-05 12:37:36 +02:00
Petr Rockai
d838501c44 toollib: Drop the bogus pv_read optimisation.
The toollib code for processing PVs given on commandline tried to be smart and
use pv_read, avoiding real metadata scanning. The premise of the optimisation
was that if there is an active MDA on the PV, it can be reliably used to
establish VG membership. This assumption is however wrong, since it's entirely
possible that this particular MDA is simply out of date, and the PV is not
actually in any VG. While this is a subtle and rare bug, it's still a bug. One
way to trigger it would be to hotplug an old disk which at some point was a PV
in an existing VG, having a seemingly valid MDA but with an old sequence
number. All commands that use normal scanning realize that this PV does not
belong into the VG anymore, but anything based on pv_read alone will believe it
does.
2013-06-05 12:37:36 +02:00
Petr Rockai
db59f5661c lvmcache: Drop a superfluous "fmt" pointer from lvmcache_info. 2013-06-05 12:37:36 +02:00
Petr Rockai
8e5d18afd3 lvmcache: Drop the label->info link. 2013-06-05 12:37:36 +02:00
Petr Rockai
ec5f5564de lvmcache: Move device ptr from lvmcache_info into label. 2013-06-05 12:37:36 +02:00
Petr Rockai
1650e1467c lvm1: Do not call into lvmcache from low-level code.
This path is not performance-sensitive anyway (LVM1 is long deprecated) and
disk-rep is very low-level code that has no business talking to lvmcache.
2013-06-05 12:37:36 +02:00
Petr Rockai
94d9562bea lvmcache: Split off lock cache into a separate unit. 2013-06-05 12:37:36 +02:00
Petr Rockai
57795df365 lvmcache: Move DA/MDA/EA lists out of lvmcache_info, into label. 2013-06-05 12:37:34 +02:00
Petr Rockai
0f49180ea7 metadata: Do not try to check vg_name of a NULL PV. 2013-06-05 12:31:18 +02:00
Zdenek Kabelac
eb2c68590b tests: bigger alignment
Test names got a bit longer...
2013-06-03 08:42:34 +02:00
Zdenek Kabelac
803b7af706 cleanup: code update and typo fix
Use the same style of loop for all filter functions.
Fix type compsoite -> Composite.
2013-06-03 08:42:33 +02:00
Zdenek Kabelac
f05c5a97c3 filters: dump filter returns error code
Add int return value from dump() function.
Report stack for error case.
Update composable filter.
2013-06-03 08:42:25 +02:00
Zdenek Kabelac
4a657a13b1 filters: compile fix
Add missing cast in previous commit.
2013-06-02 23:16:41 +02:00
Zdenek Kabelac
5467a3b2b7 filters: update composable filter
Last commit made dump filter only partially composable.
Add remaining functionality and also support composable wipe,
which is needed, when i.e. vgscan needs to remove cache.

(in release fix)
2013-06-02 22:46:06 +02:00
Zdenek Kabelac
20868482dd tests: clear write to /tmp 2013-06-02 00:50:09 +02:00
Petr Rockai
8c6f3006e7 tests: do not expose duplicate PVs in vgimportclone.sh 2013-06-02 00:50:09 +02:00
Petr Rockai
65a0650135 tests: collect all test results in files
Use files  for later processing with suffix .txt
to avoid mime type confusion.
2013-06-02 00:50:09 +02:00
Petr Rockai
7c89cbf03a tests: run all the test flavours in a single batch
This means that a test failure in one flavour no longer prevents all the
subsequent flavours from running. We also get an aggregate summary at
the end of the entire batch, instead of summaries interspersed with
progress output. Do not fail when flavour overrides are empty
2013-06-02 00:50:09 +02:00
Petr Rockai
53fbf2bea3 tests: make filter extension more robust 2013-06-02 00:50:09 +02:00
Petr Rockai
28160cb04a tests: set up global_filter instead of just filter
For testing setup global_filter.
2013-06-02 00:50:09 +02:00
Petr Rockai
e4cb88009a tests: skip the double log failure test on 2.6
On 2.6 kernels this deadlocks.
2013-06-02 00:50:09 +02:00
Petr Rockai
249ea4c34c tests: add hide_dev/unhide_dev to aux
Add helper func for easy filter adjustment.
2013-06-02 00:50:08 +02:00
Petr Rockai
1f73e992ef lvmetad: no use of persistent filter with lvmetad 2013-06-02 00:49:55 +02:00
Petr Rockai
e7878da921 filters: toplevel filter not persistent
Add a generic dump operation to filters and make the composite filter call
through to its components. Previously, when global filter was set, the code
would treat the toplevel composite filter's private area as if it belonged a
persistent filter, trying to write nonsense into a non-sensical file.
Also deal with NULL cmd->filter gracefully.
2013-06-02 00:48:58 +02:00
Petr Rockai
05bf4b8cc3 vgimportclone: override global_filter in lvm.conf
The global filter in system's lvm.conf may conflict with the custom filter we
set up in vgimportclone (they can easily fail to intersect). Since we explicitly
avoid talking to lvmetad in vgimportclone, it is safe and reasonable to do so.
2013-06-02 00:47:17 +02:00
Zdenek Kabelac
c5f7d401e5 tests: missed skip in test 2013-05-31 21:58:51 +02:00
Zdenek Kabelac
db2b65704c tests: test mirrors in clustered way
Make the clustered testing using cluster locking for most of the tests.
Use exclusive activation for mirrors and snapshot origins.
2013-05-31 21:42:32 +02:00
Zdenek Kabelac
e9e7421c8e tests: move raid test to separate file 2013-05-31 21:42:32 +02:00
Zdenek Kabelac
274eae592f tests: skip if the system already uses tested minor 2013-05-31 21:42:32 +02:00
Zdenek Kabelac
1c83c15963 tests: min snap chunk count is 3 2013-05-31 21:42:31 +02:00
Zdenek Kabelac
31f3274ed8 mirror: implement check for remotely active LV
If the mirror is active exclusively and locally, then we may proceed.
2013-05-31 21:42:31 +02:00
Zdenek Kabelac
e5baaf4ac9 lvcreate: update thin validation
Merge duplicate code that was validating lvcreate args
for creation of thin and snapshot.

Keep most of thin checks in _check_thin_parameters().

Update couple error messages.
2013-05-31 21:40:07 +02:00
Zdenek Kabelac
e01fbd9d8d lvcreate: update snapshot test
With last commit, there has been missing test for valid setting
of extents - there is no need to test max size when creating
thin snapshots.
2013-05-31 21:30:44 +02:00
Jonathan Brassow
562c678ee2 DM RAID: Add ability to throttle sync operations for RAID LVs.
This patch adds the ability to set the minimum and maximum I/O rate for
sync operations in RAID LVs.  The options are available for 'lvcreate' and
'lvchange' and are as follows:
  --minrecoveryrate <Rate> [bBsSkKmMgG]
  --maxrecoveryrate <Rate> [bBsSkKmMgG]
The rate is specified in size/sec/device.  If a suffix is not given,
kiB/sec/device is assumed.  Setting the rate to 0 removes the preference.
2013-05-31 11:25:52 -05:00
Jonathan Brassow
7061b74ec6 Clean-up: Fix copy-n-paste error
Description for --with-raid should be s/mirror/raid/.
2013-05-31 11:14:37 -05:00
Zdenek Kabelac
09cc5f65db tests: snapshot needs 3 chunks 2013-05-30 17:56:47 +02:00
Zdenek Kabelac
6f9366696c cleanup: fix tabs 2013-05-30 17:56:41 +02:00
Zdenek Kabelac
a766ea16b8 tests: snapshot testing 2013-05-30 17:35:23 +02:00
Zdenek Kabelac
cd9350198e tests: drop testing of raids in cluster
Since raid is not clustered yet - drop it from testing in cluster phase.
(saving running some time)
2013-05-30 17:35:23 +02:00
Zdenek Kabelac
e4dfa785d1 libdm: compensate suspend counter for live table
This patch may not be fully correct. It tries to solve
the imbalanced suspend counter.

The problem starts when some LV is created and fails in resume path.
(i.e. resuming to large PV (enforced) over small loop devices)

This fails in _resume_node() after dm_task_run(). And while
existing device with empty table is left in inactive table,
further calls are reporting this device is in suspend state.

When later the lvm2 tries to rollback created device and deactivate it,
it will end with internal error, when we try to decrement
never incremented suspend counter.

As an 'easy fix' for now update suspend counter only for live nodes.

TODO: explore better fix.
2013-05-30 17:35:23 +02:00
Zdenek Kabelac
b08c346489 lvresize: drop monitoring of 100% covered origin 2013-05-30 17:35:23 +02:00
Zdenek Kabelac
3ced1bf694 lvresize: check for max snapshot size
As for lvcreate, lvresize also doesn't need to grow bigger then needed.
2013-05-30 17:35:23 +02:00
Zdenek Kabelac
bd3ece0128 lvcreate: reduce too large cow
Detect maximum usable size of snapshot COW device,
and do not waste more space for such LV then needed.
2013-05-30 17:35:14 +02:00
Zdenek Kabelac
eb7e206a73 snapshot: add cow_max_extents
Add more precise calculation of the maximum usable snapshot size.
Using only percentage fails for small size of snapshot and extents.
2013-05-30 17:30:15 +02:00
Zdenek Kabelac
59962d8d3e snapshot: require 3 chunks for creation
There is no point in creation of 2chunks snapshot,
since the snapshot is invalidated immeditelly with the first write
as there is no free chunk for COW blocks
(2 chunks are used by the snap header and the 1st. metadata chunk).

Enhance error message about the lowest usable size.
2013-05-30 17:28:03 +02:00
Zdenek Kabelac
39705ed201 cleanup: test for -1 value for lseek error
Error is defined as -1.
Yet it seems we are unable to seek to 0 for devices > 8Exabytes.
2013-05-30 17:26:42 +02:00
Zdenek Kabelac
56779c32c5 snapshot: fix resize of 100% full cow
When the COW area is using all the available space (100%) it can be still
a valid snapshot which may need a resize. So support it.
2013-05-30 17:26:20 +02:00
Zdenek Kabelac
99f0483580 args: do not accept >=16EiB sizes
Instead of seeing wierd overflows inside the lvm code,
giving false error messages, kill the user experiment in the begining.

Who needs to use more then 16EiB with lvm2 and 64bit anyway...
2013-05-30 17:23:51 +02:00
Zdenek Kabelac
c5c452a137 libdm: add public macro DM_TO_STRING
Since this stringifying macro is usable in more places,
move it to the public header with DM_ prefix.
2013-05-30 16:53:59 +02:00