Tony Asleson [Tue, 27 Sep 2016 03:02:08 +0000 (22:02 -0500)]
lvmdbusd: Allow PV device names to be '[unknown]'
When a PV device is missing lvm will return '[unknown]' for the device
path. The object manager keeps a hash table lookup for uuid and for PV's
device name. When we had multiple PVs with the same device path we
we only had 1 key in the table for the lvm id (device path). This caused
a problem when the PV device transitioned from '[unknown]' to known as any
subsequent transitions would cause an exception:
Traceback (most recent call last):
File "/usr/lib/python3.5/site-packages/lvmdbusd/request.py", line 66, in run_cmd
result = self.method(*self.arguments)
File "/usr/lib/python3.5/site-packages/lvmdbusd/manager.py", line 205, in _pv_scan
cfg.load()
File "/usr/lib/python3.5/site-packages/lvmdbusd/fetch.py", line 24, in load
cache_refresh=False)[1]
File "/usr/lib/python3.5/site-packages/lvmdbusd/pv.py", line 48, in load_pvs
emit_signal, cache_refresh)
File "/usr/lib/python3.5/site-packages/lvmdbusd/loader.py", line 80, in common
cfg.om.remove_object(cfg.om.get_object_by_path(k), True)
File "/usr/lib/python3.5/site-packages/lvmdbusd/objectmanager.py", line 153, in remove_object
self._lookup_remove(path)
File "/usr/lib/python3.5/site-packages/lvmdbusd/objectmanager.py", line 97, in _lookup_remove
del self._id_to_object_path[lvm_id]
KeyError: '[unknown]'
when trying to delete a key that wasn't present. In this case we don't add a
lookup key for the device path and the PV can only be located by UUID.
The dm_stats_delete_region() call removes a region from the bound
device, and, if the region is grouped, from the group leader
group descriptor stored in aux_data.
To do this requires a listed handle: previous versions of the
library do not since no dependencies exist between regions without
grouping.
This leads to strange behaviour when a command built against an old
version of the library is used with one supporting groups. Deleting
a region with dmstats succeeds, but logs errors:
# dmstats list
Name RgID RgSta RgSiz #Areas ArSize ProgID
vg_hex-root 0 0 1.00g 1 1.00g dmstats
vg_hex-root 1 1.00g 1.00g 1 1.00g dmstats
vg_hex-root 2 2.00g 1.00g 1 1.00g dmstats
# dmstats delete --regionid 2 vg_hex/root
Region ID 2 does not exist
Could not delete statistics region.
Command failed
# dmstats list
Name RgID RgSta RgSiz #Areas ArSize ProgID
vg_hex-root 0 0 1.00g 1 1.00g dmstats
vg_hex-root 1 1.00g 1.00g 1 1.00g dmstats
This happens because the call to dm_stats_delete_region() is inside
a dm_stats_walk_*() iterator: upon entry to the call, the iterator
is at its end conditions and about to terminate. Due to the call to
dm_stats_list() inside the function, it returns with an iterator at
the beginning of a walk and performs a further iteration before
exiting. This final loop makes a further attempt to delete the
(already deleted) region, leading to the confusing error messages.
libdm: fix stats walk compatibility with older dmsetup
The current dmsetup.c handles DR_STATS and DR_STATS_META reports
separately in _display_info_cols(), meaning that the stats walk
functions are never called for these report types.
Versions before v2.02.159 have a loop using dm_stats_walk_do() and
dm_stats_walk_while(), that executes once for non-stats reports,
and once per region, or area, for DR_STATS/DR_STATS_META reports.
This older behaviour relies on the documented behaviour that the
walk functions will accept a NULL pointer as the struct dm_stats*
argument.
This was broken by commit f1f2df7b: the NULL test on dms and
dms->regions were incorrectly moved from the dm_stats_walk_end()
wrapper to the internal '_stats_walk_end()' helper.
Since the pointer is dereferenced in between these points, using
an older dmsetup with current libdm results in a segfault when
running a non-stats report:
# dmsetup info -c vg00/lvol0
Segmentation fault (core dumped)
Restore the NULL checks to the wrapper function as intended.
Peter Rajnoha [Tue, 27 Sep 2016 08:47:54 +0000 (10:47 +0200)]
systemd: disable service start rate limiting for lvm2-pvscan@.service
We shouldn't be losing pvscans just because of the fact that the
underlying device (PV) appears and disappears quickly in the system,
otherwise lvmetad may not see the device if it appears again (or it may
still keep the device in cache even it's already gone).
Peter Rajnoha [Fri, 23 Sep 2016 12:51:15 +0000 (14:51 +0200)]
toolcontext: read all configuration sources when checking config values in lvm2-activation-generator through lighweight toolcontext handler
We added lightweight toolcontext handle to avoid useless initialization
of some parts of the context and also to avoid problems when using the
handle very soon at system boot, like in lvm2-activation-generator
through lvm2app interface. However, we missed reading all the other
config sources like lvmlocal.conf as well as any tag config - we need to
read these too to get the final config value which may be overriden in
any of these additional config sources.
Currently, we use this lightweight toolcontext handle to read
global/use_lvmetad and global/use_lvmpolld config values in
lvm2-activation-generator using lvm2app interface (lvm_config_find_bool
lvm2app function).
tests: fix raid rebuild tests to work with older target versions
Pre 1.9 dm-raid targets status output was racy, which caused
the device status chars to be unreliable _during_ synchronization.
This shows paritcularly with tiny test devices used.
Enhance lvchange-rebuild-raid.sh to not check status
chars _during_ synchronization. Just check afterwards.
dmsetup: ensure --filemap histogram bounds are freed
Make sure that the temporary dm_histogram used for the bounds
argument is freed in the case that the user provided a --bounds
argument on the command line.
The dm-raid target now rejects device rebuild requests during ongoing
resynchronization thus causing 'lvconvert --repair ...' to fail with
a kernel error message. This regresses with respect to failing automatic
repair via the dmeventd RAID plugin in case raid_fault_policy="allocate"
is configured in lvm.conf as well.
Previously allowing such repair request required cancelling the
resynchronization of any still accessible DataLVs, hence reasoning
potential data loss.
Patch allows the resynchronization of still accessible DataLVs to
finish up by rejecting any 'lvconvert --repair ...'.
It enhances the dmeventd RAID plugin to be able to automatically repair
by postponing the repair after synchronization ended.
More tests are added to lvconvert-rebuild-raid.sh to cover single
and multiple DataLV failure cases for the different RAID levels.
Commit 199697accff0658a11420e263c1f85fb74cd39d3 rerouted funtion
for priting cache volume origin to lvm2app app function - which
however had a bug. So restore the original functionality
and print correct LV as cache origin LV.
Tony Asleson [Mon, 19 Sep 2016 14:54:24 +0000 (09:54 -0500)]
lvmdbusd: Fix dbus object with only properties and no method
Gris debugged that when we don't have a method the introspection
data is missing the interface itself eg.
<interface name="<your_obj_iface_name>" />
When adding the properties to the dbus object introspection we will
add the interface too if it's missing. This now allows us the
ability to have a dbus object with only properties.
Tony Asleson [Sat, 17 Sep 2016 05:11:59 +0000 (00:11 -0500)]
lvmdbustest.py: Test with/without timeout values
Because of the different code paths we need to test job handling with
all operations. Test now runs virtually everything with timeout == 0
and timeout == 15 so that we test both use cases.
Note: If a client passes -1 for the timeout value they need to make
sure that their connection time out is also infinite. Otherwise, if the client
times out the service side hangs any new dbus calls until the job
that is in progress completes. Not sure why this behavior is occuring
at this time, but it appears a limitation/bug of the dbus-python library.
Tony Asleson [Sat, 17 Sep 2016 05:10:46 +0000 (00:10 -0500)]
lvmdbusd: Set value to non none on error
When we register a failure we need to use a valid value which will be
returned with the object manager. Otherwise we will raise an Exception
because we are trying to construct an object path from None.
Tony Asleson [Fri, 16 Sep 2016 19:00:14 +0000 (14:00 -0500)]
lvmdbusd: Vg.LvCreate* methods, correct return type
The methods were returning an instance of the object instead of the
object path which was causing an exception when the result was returned
with the job object as we are explicity trying to return an object path.
Unit test added which re-creates the issue and verifies the fix.
Reload of thin-pool origin_only is designed to only post messages
to a thin-pool. It's not intended to be used for reload of thin-pool
table. Fix it by using standard call 'lv_update_and_reload()'.
thin: enforce there is some free space in thin pool metadata
Unconditionally guard there is at least 1/4 of metadata volume
free (<16Mib) or 4MiB - whichever value is smaller.
In case there is not enough free space do not let operation proceed and
recommend thin-pool metadata resize (in case user has not
enabled autoresize, manual 'lvextend --poolmetadatasize' is needed).
thin: report pool as holder when no active thin volume
In the case there is no active thin volume, report thin pool
as lock holder. This fixed function like lvextend
which either expecte lock holder LV is some active thin
or 'possibly' inactive thin pool.
David Teigland [Tue, 13 Sep 2016 21:50:47 +0000 (16:50 -0500)]
lvmlockd: fix metadata validation when rescanning VG
When rescanning a VG from disk, the metadata read from
each PV was compared as a sanity check. The comparison
is done by exporting the vg metadata from each dev to
a config tree, and then comparing the config trees.
The function to create the config tree inserts
extraneous information along with the actual VG metadata.
This extra info includes creation_time. The config
trees for two devs can easily be created one second
apart in which case the different creation_times would
cause the metadata comparison to fail. The fix is to
exclude the extraneous info from the metadata comparison.
Use issue_discard to lower memory demands on discardable test devices
Use large devices directly through prepare_pvs
I'm still observing more then 0.5G of data usage through.
Particullary:
'lvcreate' followed by 'lvconvert' (which doesn't yet support --nosync
option) is quite demanging, and resume returns quite 'late' when
a lot of data has been already written on PV.
Reinstantiate reporting of metadata percent usage for cache volumes.
Also show the same percentage with hidden cache-pool LV.
This regression was caused by optimization for a single-ioctl in
2.02.155.
cache: scrubbing for cache origin LV - Bug 1169495
Allow RAID scrubbing on cache origin sub-LV
This patch adds the ability to perform RAID scrubbing on the cache
origin sub-LV (https://bugzilla.redhat.com/1169495). Cache origin
operations are restricted to non-clustered RAID LVs until there can
be further testing in a cluster (even for exclusive activation).
User can either specify directly _corig LV
or he can specify cache LV and operation --syncation is
passed ONLY to _corig LV.
If users wants to manipulation with cache-pool devices - he
needs to specify this object name.
Signed-off-by: Jonathan Brassow <jbrassow@redhat.com>
Peter Rajnoha [Tue, 6 Sep 2016 11:12:02 +0000 (13:12 +0200)]
dev-type: check for DEVLINKS udev db variable existence if udev_device_get_is_initialized fn is not present
Older udev versions (udev < v165), don't have the official
udev_device_get_is_initialized function available to query for
device initialization state in udev database. Also, devices don't
have USEC_INITIALIZED udev db variable set - this is bound to the
udev_device_get_is_initialized fn functionality.
In this case, check for "DEVLINKS" variable instead - all block devices
have at least one symlink set for the node (the "/dev/block/<major:minor>".
This symlink is set by default basic udev rules provided by udev directly.
We'll use this as an alternative for the check that initial udev
processing for a device has already finished.
Peter Rajnoha [Mon, 5 Sep 2016 09:33:08 +0000 (11:33 +0200)]
lvmetad: check udev for mpath component several times if udev record not initialized yet
It's possible (mainly during boot) that udev has not finished
processing the device and hence the udev database record for that
device is still marked as uninitialized when we're trying to look
at it as part of multipath component check in pvscan --cache code.
So check several times with a short delay to wait for the udev db
record to be initialized before giving up completely.
David Teigland [Wed, 31 Aug 2016 18:05:53 +0000 (13:05 -0500)]
lvmetad: use udev to ignore multipath components during scan
When scanning devs to populate lvmetad during system startup,
filter-mpath with native sysfs multipath component detection
may not detect that a dev is multipath component. This is
because the multipath devices may not be set up yet.
Because of this, pvscan will scan multipath components during
startup, will see them as duplicate PVs, and will disable
lvmetad. This will leave lvmetad disabled on systems using
multipath, unless something or someone runs pvscan --cache
to rescan.
To avoid this problem, the code that is scanning devices to
populate lvmetad will now check the udev db to see if a
dev is a multipath component that should be skipped.
(This may not be perfect due to inherent udev races, but will
cover most cases and will be at least as good as it's ever
been.)
Peter Rajnoha [Tue, 30 Aug 2016 13:33:50 +0000 (15:33 +0200)]
lvmdump: use lsblk -s and lsblk -O in lvmdump only if these options are supported
The lsblk is just a nice helper here - it's not crucial for lvmdump so
do best effort here and use the most we can from current version of
lsblk that is installed on system. The lsblk -s option was added a bit
later after lsblk introduction and lsblk -O support even more later -
so if these are not available, use only pure lsblk output without any
extras.