For some reason the leaded descriptor warning message coming
from LVM is causing Cinder to fail startup and it appears to be
masking out the vg response in vgs calls.
We typically don't hit this, but due to the nature of Kolla and
I guess going through the different processes via the containers
this gets logged every time vgs is called. Eric Harney rightly
pointed out that rather than use exception handling and such
that we should use the LVM env variable mechanism we already have
in place in Cinder.
For now this patch added a new config option to the LVM driver:
lvm_suppress_fd_warnings=True|False
This is useful for K8 deployments that have an indirect call to the
LVM cmds which results in failures.
For those that are interested, this can also be done outside of
cinder by setting the silence_logs variable in lvm.conf
This is made optional as a config flag to avoid any breakage for
existing deployments during upgrade.
DocImpact: Need to add info regarding config update to docs
Change-Id: I85612fa49475beea58d30330c8fe8352a2f91123
Closes-Bug: #1619701
(cherry picked from commit 055ec1ce73)
The source volume doesn't get deleted at the end of the migration.
This is due to volume ID DB swap. Solved by using name_id instead of id.
Fixed test to use fake_volume_obj
Change-Id: I45b8bc8304b51346c1fcc7c5a673d928281a1eb4
Closes-Bug: #1618700
(cherry picked from commit a18253fa49)
Turns out I7dae83667a1aca92c552fbfaa1e90c6558e293bf had a mistake and
was unconditionally converting RequestSpec o.vo to dictionary just
before sending it to the scheduler. This wasn't noticed as scheduler
manager has compatibility shim in place that converts the dictionary
back into o.vo when it's received.
The problem this produces is that we cannot drop the compatibility
shim from manager until 4.0 scheduler RPC API. This commit fixes that.
Change-Id: I7beaaddcff079dfe7a0032e2cf2afa75a5e88522
Closes-Bug: 1625170
(cherry picked from commit 32bfddf841)
It was found that a modified or corrupted image file can cause a DoS
on the host when getting image info with qemu-img.
This uses the newer 'prlimit' parameter for oslo.concurrency execute
to set an address space limit of 1GB and CPU time limit of 2 seconds
when running the qemu-img info command.
Change-Id: If5b7129b266ef065642bc7898ce9dcf93722a053
Closes-bug: #1449062
(cherry picked from commit 78f17f0ad7)
Fixed miscellaneous errors dealing with the state of Dell's
Live Volumes after a LV AFO event.
Was looking for the original volume and sc server object
after the primary was down. We now look for the new primary
and attach that.
Ensure export was failing to properly kick off a swap
back to the original primary if it was available.
Fixed an error with deleting replications.
Added a workaround to finding a ScLiveVolume object. Previous
method could return stale information.
Change-Id: I58889adbcf279d4b0107cbc51e1dace888a3fb15
Closes-Bug: #1617401
(cherry picked from commit 45bc8abc47)
Conditional update method from the persistent OVO class is blindly
relying on persistent OVOs having an OPTIONAl_FIELDS attribute without
defining a default.
This patch adds an empty default so we can do conditional updates on
persistent OVO instances that don't have optional fields.
Change-Id: Icf640da34df0990b5ad2609d5d230ac9b0a51311
(cherry picked from commit 2987b33970)
Current code has OPTIONAL_FIELDS as part of the VolumeType library
instead of as a class attribute of the VolumeType OVO class. This is
not how it should be, as there are methods -like conditional update-
that rely on the OPTIONAL_FIELDS being there.
This patch moves the OPTIONAL_FIELDS inside the VolumeType class.
Change-Id: I18404a2f32abc6c9d9b7d932a7acd299c4e708e5
(cherry picked from commit f63f1463c6)
In Mitaka as a workaround for creating volumes without volume type c-api
was sending an empty dictionary as request_spec['volume_type'] in
create_volume() RPC cast to scheduler. This changed in Newton and we're
handling this situation directly in scheduler, but when running Newton's
c-api with Mitaka's c-sch we should mimic older behavior. This patch
implements that.
Change-Id: I3dbb1934f3d8e68fef49f56155cbf59bc5fadc3a
Closes-Bug: 1619008
(cherry picked from commit 91bdf9de4c)
The VolumeV3Proxy method was calling the manager
with an extra parameter. This removes it.
Change-Id: I8471ea9649a2c238aac9c2bd0ff7cd995ef7ffef
Closes-Bug: #1626351
(cherry picked from commit 051ff39a9c)
The driver could return duplicate iSCSI portal/IQN/Lun combinations.
This patch returns the condensed list.
Also sometimes when Cinder attempts to attach to a volume it will use
the first item in the multipath list rather than the specified single
path entries. So this patch simply moves a known up and active entry
to the front of the list.
Change-Id: Idb371825b6520d4bdb09a1dbb37429cb10643e93
Closes-Bug: #1622703
(cherry picked from commit 4708aa16a7)
user_id, project_id and status were listed as request parameters when
they are actually response parameters.
Change-Id: I5ff464c6aae853929b2b10ba62898913977016b0
Closes-Bug: 1603177
This patch creates volume RPC API version 3.0, while retaining
compatibility in rpcapi and manager for 2.x, allowing for continuous
deployment scenarios.
This should be merged just before the Newton release.
Change-Id: Ib927ccb3bb1f91e441f5b8540a3e39acefa4a6ab
This patch creates scheduler RPC API version 3.0, while retaining
compatibility in rpcapi and manager for 2.x, allowing for continuous
deployment scenarios.
This should be merged just before the Newton release.
Change-Id: I8eb9c1ae93d84c63a061294fc570be1d9eed69ba
Currently, the API reference is showing a blank sample response
for the v2 "List API Versions" call [1]. That appears to be because
the sample file it includes is "versions-resp.json" which exists, but
is 0 bytes (completely empty). There is also a
"versions-response.json" file present in the samples directory, which
appears to contain a valid JSON reply.
This patch fixes the problem by removing the 0-byte sample file and
pointing to the correct one.
[1]
http://developer.openstack.org/api-ref/block-storage/v2/#list-api-versions
Change-Id: If2c1b486229681f68ba4ab0c13307de4599d9b72
Closes-Bug: #1623744