Age | Commit message (Collapse) | Author | Files | Lines |
|
Swift proxy-servers use memcache to store and share metadata. This
change adds swift.proxy-memcache metadata to the swift-source and
swift-storage-source yaml templates modelled on the existing
swift.devices metadata. This metadata will be consumed by the
swift-proxy/os-config-applier/etc/swift/proxy-server.conf element
if the metadata exists.
Change-Id: If0b5724f69e7ec1c98e4dbdbeb9f08c4a18151b6
|
|
-adding Undercloud Ceilometer that will collect statistics
about Overcloud nodes, via SNMP
Change-Id: I1e90ad8d5bad16bc1c418ca2dbd78163abe6267c
|
|
This migrates the overcloud to using OS::Heat::StructuredConfig and
OS::Heat::StructuredDeployment. With those tools, we can decouple
servers from software configuration and begin to deprecate features in
tripleo_heat_merge.
Change-Id: Ice85f0711e90d0fabf1d1bc4698201c4d6758508
|
|
|
|
|
|
Updates all references for notCompute and notcompute
to use 'controller' instead.
Change-Id: I70ef83f35064ab388bdc7e1a6da62b6585580010
Partial-bug: #1300324
|
|
|
|
|
|
|
|
|
|
|
|
Change-Id: I30c0e175fda448a300cae0b233757d31ce73402f
|
|
Add the enable_tunneling option to the neutron-openvswitch-agent
metadata for the swift storage template. If not preset, this option is
left blank in the openvswitch plugin configuration plugin. This causes
the service to fail to start because a blank value is not permissible
for a boolean option.
Change-Id: Ieadcdab295913121bd00dbd25e4245024bc2240f
|
|
Innodb buffer pool size in mysql config file is now set to 592M,
this value is for VMs with 2G RAM too much. This patch makes the
pool size configurable and decreases default value to 100 MB
(this should be sufficient for default devtest setup). A patch which
for tripleo-image-element which sets this value in my.cnf will follow.
Change-Id: I0d2529abe51d4dc34415bf6c40da7a267bf3e65b
|
|
This is necessary to support iscsi helpers other than tgtadm, such
as lioadm.
Change-Id: I8a63872185f73372fcf6c90d3dde31e5bb0a168b
|
|
Add a BlockStorage0Config resource in the block-storage.yaml template.
This was missing previously and was causing the behavior of having all
block storage nodes only being created after all other nodes were up in
a deployment (I presume b/c of the wait conditions).
Also, we need the BlockStorage0Config resource so that we can get the
configuration of neutron-openvswitch-agent from the metadata, previously
the values in the file were unset, causing the service to fail to start.
Change-Id: I297de24079d1ece66d35213b4ef1f2c8c50e73f0
|
|
We don't make use of dnsmasq_range anywhere so lets remove it.
Change-Id: Id3e14aee8a9f720a09bc2c8ace99e06c4a2d09a9
|
|
Updated heat templates to default to preserve ephemeral partition
on a heat stack update. Allows undercloud and overcloud to be
re-imaged/updated and state preserved.
Change-Id: I6626af48d8c55672022a46fe48e5725ad2619f0c
|
|
|
|
|
|
The merge of ironic-source wasn't actually merging, leading to bad
templates. As part of this add the needed SSH virtual power key
(because like nova-bm, Ironic requires that for now).
We also need to set the compute_manager for nova with Ironic, so that
gets set too.
Change-Id: I0d7d535dbac3584b3eb12865259ee6403e20c98f
|
|
Capture some undercloud metadata into the overcloud compute
configs, so the overcloud nodes can pull updated images from
the undercloud glance. Unset by default, but able to be set
during stack-create or stack-update.
Change-Id: I2997d23c584055c40034827e9beb58e6542ea11c
|
|
Allows image policy to be specified when updating the undercloud
stack. This enables the preservation of persistent ephemeral disks.
Change-Id: Id7bc0a72ae1737d2fdadb3f47006bd0cfb0bc42f
Closes-Bug: #1295616
|
|
|
|
This patch changes the location of the Ironic Nova compute driver
in nova.conf so that it uses the driver from within the Ironic
directory tree. This patch will need to be reverted once the Ironic Nova
driver lands in the nova.
Change-Id: I9199d506f3665cb0827edd642a1806713a42b90e
|
|
Previously all cinder functionality was in a single element and
required "volume: 'true'" to install the cinder-volume service.
Now cinder has been split into multiple elements and you build
an image with the cinder-volume element to achieve the same goal.
This patch removes the "volume: 'true'".
Change-Id: I2ed2e6f7c7db7619aee7ce4ba730f310344e0b0e
|
|
In some environments the heat stack-create command would complete
even though the block storage node would still be in a
build/spawning state. This patch adds a wait condition to ensure
the block storage node has a chance to complete and the stack-create
command to complete properly.
Change-Id: Idb524625b8d6df1ff6f927c22b72bc6d3d611686
Partial-Bug: #1290764
|
|
|
|
|
|
|
|
This new parameter will be used to configure Heat for the new required
stack domain admin.
Change-Id: Ieace33154a14f472aab8ac76b2b876db77230110
Partial-Bug: #1287453
|
|
|
|
This patch removes the duplicate rabbit username and password
parameters. These parameters cause a conflict when running
merge.py because they already exist in the overcloud template.
Change-Id: I01d386577734c7c516a477e1e402882294ac4376
|
|
This patch removes the seperate overcloud-with-block-storage.yaml
make target and merges the functionality into the main overcloud.yaml
target. The default BLOCKSTORAGESCALE is 0. Change this number to add
additional block storage nodes.
This patch also updates the overcloud-with-block-storage-nfs target
to more closely resemble the overcloud.yaml target, including swift
and ssl templates as needed. The default BLOCKSTORAGESCALE is 1.
Change this number to add additional block storage nodes.
Change-Id: Ifcc733cfbf9f356393c3f6c72b2bbbb0214d3040
|
|
|
|
Updated the existing swift-source template to properly list all
swift devices.
Created a new swift-storage-source template to add additional
swift-storage only nodes.
The default SWIFTSTORAGESCALE is 0. Change this number to add
additional swift storage nodes.
Change-Id: Ia05f4ad44593316430d60541106d23b032c3e760
|
|
To support the changes to the cinder element to allow the nfs
backend to be utilized, this template has been added to show
the usage of the nfs_shares key in the cinder metadata.
The value is a list of strings containing share addresses.
This change is added to facilitate an example to test
the cinder element change: https://review.openstack.org/#/c/74563/
You may setup the nfs server manually or use the nfs-server
element at https://review.openstack.org/#/c/74712/.
Change-Id: I5b6cb118b34421ea07a81ed1fe68db24b1d4f19d
|
|
|
|
|
|
Lines for notifier-strategy and log-file had 1 level too much of
identation causing an error in the available metadata.
Change-Id: I559855887349bf2bbab87788ef9baa216bde60d8
|
|
When Merge::Map was used like so:
NovaCompute0Config:
Merge::Map:
NovaCompute0:
Fn:....
We were adjusting the inner NovaCompute0 to the current scaling loop
position rather than exploding it out as part of the map. For now, we
want maps to be global in context so the fix is fortunately simple.
Change-Id: Ibea723d4c763f593d934a9d2acfd31ce369f6408
|
|
Just developer convenience - I'm used to autotools 'make check'
environments.
Change-Id: Idf5bb882f798d22d4513756b5273d4ae19caa5c0
|
|
|
|
|
|
Currently our wait conditions are racing with Heat resolving
configurations. There should be plenty of time but sometimes Heat may
be dealing with a temporarily problematic Nova API and spinning on that.
While that is happening, the in-instance tools will not have their full
configuration available to them. We don't want the wait condition timeout
to start until the box has had its actual config exposed to it.
Change-Id: I0eab8fe7547d3cbcebb1559fd3d06206b1750e96
|
|
Expose 2 new parameters to control glance configuration:
- GlanceLogFile
- GlanceNotifierStrategy
This will allow the glance configuration options log_file and
notifier_strategy to be set in glance-api.conf and glance-registry.conf
Change-Id: I1af1ecfdfe0859f2baba4579aab6e947a2a54308
|
|
All nodes in rabbitmq cluster should have same cookie. This patch adds
default random cookie to existing templates.
Change-Id: I6579089bd95b0169db20635479e077f3ecade8e2
|
|
- different Flavors for control, compute and storage nodes
- for devtest use, they default to 'baremetal', so nothing
changes
- for Tuskar, there is a possibility to have them different for
every role
Change-Id: I8c1b80f55a91c7a7fd5e560ccdb8da82ec374084
|
|
Not '.local'. Doh.
Change-Id: I61ac6e5d6dfefcfe8c78618f6fc5d4c2c61831f0
|
|
Nova compute uses hostnames to do live migration, so we need this, or
DNS, and for now this is simpler, as we don't have Designate available
in the undercloud.
Change-Id: Idf0b106c5fbe60680b23a730eb37619a3d796869
|