Age | Commit message (Collapse) | Author | Files | Lines |
|
JIRA: APEX-82
Change-Id: I0f2164a67b425ae5c99cd6bfee2182b23867b4cf
|
|
|
|
JIRA: APEX-74
/var/log/messages for each node will be slurped into a new folder in the
stack user's home directory. If debug was passed, it will also be printed
out along with the nova list for that node.
Change-Id: I726da67b076bdf1fb38ba1805456cbd7c7d9e096
|
|
Change-Id: I3474118e58ae4946abfe9e74bb73084aad3634b2
JIRA: APEX-88
|
|
- Forcing no-ha to 1 compute node
- updating install-packages to allow detached deploy
- updating kernel to one with a growfs bug fix
- removing service script with growfs work around
- adding rootpw to overcloud when in debug mode
JIRA: APEX-85, APEX-76, APEX-73
Change-Id: I202275c9cbeb864a044b119a9e440d59a1955c40
|
|
This will persist networks across reboots instead of creating a
transient network.
Change-Id: I98b8e0c577c145197dd7195274b27ded2cc96297
|
|
|
|
Change-Id: I6c8bd9a949b1b818bedd4994f490da4a703ee6a5
Signed-off-by: Ryota MIBU <r-mibu@cq.jp.nec.com>
|
|
Note this image could change from opendaylight to a regular overcloud
image if we choose to split out a nosdn rpm in the future. For now
fallback to the ODL image.
JIRA: APEX-78
Change-Id: I959d917330b1d8f68c16c3d122901494112db071
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Fixes include:
- sfc deployments now are non-HA, and disable swift and ringbuilder as
it causes issues with the 3.13 kernel
- ODL Features boot is now parameterized in heat templates
- SFC features boot is now set to "odl-ovsdb-sfc-rest"
- Coexistence table offsets included if SFC deployment
- VXLAN workaround applied for 123.123.123.x subnet if SFC
Change-Id: Ide5e0b451adcb98c417e57628a51302f3123647f
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Daily failed because, we removed the default value of "TRUE" for
ha_enabled, and deploy settings sets it as "true", which we were not
looking for as a boolean in deploy. This patch fixes that.
Change-Id: I03ac599aa2a35ae494e13f9293aef273ab3051e9
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
|
|
Changes Include:
- New Interactive mode. Allows you to pause deployments in order to
make changes before continuing (or exit and deploy manually). Useful
when needing to modify NIC order for baremetal deployments.
- deploy_settings -d is now required.
- ha_enabled no longer defaults to true. It can be specified in
deploy_settings or overridden with via command line.
- compute nodes now scale correctly with what is defined by inventory
JIRA: APEX-70 APEX-71 APEX-72
Change-Id: Id45808f4d0435d66f12c020bd4455faaa43da191
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
URI was being passed incorrectly by tht
JIRA: APEX-59
Change-Id: If9912f1ef0618836f37931e3573293d0f434d23f
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Changes Include:
- Disables introspection. We don't currently use introspection for
anything other than to just run it. Disabling it to increase deploy
speed and reduce pxeboot issues.
- Adds timeout to deploy of 90 minutes. Default heat timeout is 3-4
hours.
- Fixes OVS bridge attach/detach for users who use PREFIX in their
ifcfg instead of NETMASK
- Adds nosdn deploy settings file, and allows nosdn deployments
- Adds example inventory file to the docs dir
Change-Id: Ic13d6c4d5f27bac25365e5951acdd9bd3e641a15
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
ONOS needs the capability to have br-ex on compute nodes with no IP for
L3 functionality to work correctly on baremetal. This patch allows that
capability, and makes it default behavior (sdn_l3 true). If desired, a
user can set sdn_l3 to false, and compute nodes public interface will be
a regular interface with no IP for ONOS deployments.
JIRA: APEX-64
Change-Id: I578b946edbddf7fc9c29363c78fec2c3ada48cf6
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Note this behavior has changed in TripleO post Brahmaputra. The
endpoints and services are created in THT rather than hardcoded in
tripleoclient.
JIRA: APEX-65
Change-Id: Ie595f2778fb509213e2992bf1ef60df82cf22666
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
ONOS public network needs to not have IPs on the compute nodes' public
interface.
JIRA: APEX-64
Change-Id: I0d4d97b1b7dcaa39395940da03ae6b19f1dd9eb0
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
JIRA: APEX-62
Change-Id: Ide34d9971cd6ac987b5994f5e23b4cfa58f92b29
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
JIRA: APEX-63
Change-Id: I4155111eb2c19d74e0f265ccac25756f35bb41a0
|
|
JIRA: APEX-46
JIRA: APEX-48
Change-Id: Ic2a0d9035429c5887d10f60febacff63dd2eeb5e
|
|
Fixes:
- opendaylight was not being installed on controllers due to missing
param
- controller nics were not being found due to using computes extension
of br-ex for ODL L3 (controllers have br-ex by default)
Change-Id: Ic1b6d57fe162194030607856120a742dbed1e9e2
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Adds support for ODL L3 which uses DVR (distributed virtual routing).
Due to DVR compute nodes must have br-ex. In addition, neutron must
force metadata via dhcp agent, since there is no neutron L3 agent
running. ODL L3 runs as custom plugin to neutron.
Also adds support for configuring IP on public NIC now by default. This
is needed for ONOS deployment, and access via public interface.
Change-Id: I5df0e49bd903efacf8c87486da27576b01ce5779
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Change-Id: I47c316e26ff8e597c781562645397335e8c5bd70
Signed-off-by: bob zhou <bob.zh@huawei.com>
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Changes include:
- Scenarios for opencontrail, onos, sfc, odl l2/l3
- Minor fix to error print statement
- Post Config is now disabled when using --flat
Change-Id: I5f3686dda4c4a573b1777916669053593852676a
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
In a local test, I observed virt-customize failing during ssh
key injection
Injecting ssh key to instack VM
[ 0.0] Examining the guest ...
[ 48.0] Setting a random seed
[ 48.0] Running: mkdir /root/.ssh/
mkdir: cannot create directory '/root/.ssh/': File exists
virt-customize: mkdir /root/.ssh/: command exited with an error
Adding -p should prevent this from happening.
Change-Id: If8f085cba3083ccf3d5e8f648a9830f142ea53d9
|
|
We should only bridge to admin/public for baremetal deployments. There
is no reason for us to bridge to the private or storage networks on
instack for deployment.
Also fixes OpenDaylight to Lithium 3.3
Change-Id: I2eb8f83ddbcb69d3219e7927025b46dcb9550ba9
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
- adding a missing exit 1 in the deploy script
- setting link up on hypervisor bridges
- redirecting error from cleanup commands so they're not so chatty
Change-Id: I844c35effb2817948ea096b07a34d83678803e00
|
|
|
|
Fixes issues with common-functions:
- ipcalc in epel is old and does not support HOSTMIN,HOSTMAX. Now
BROADCAST and NETWORK values used to find first and last IP in
subnet.
- fixes to auto-generation where values conflicted for IP ranges
- provisioner_ip now generated correctly
Change-Id: I03c7f841d2e1edb1a50041f53970318b6a718cd9
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Neutron external net should be creatd on overcloud
Change-Id: I2c845e9af40d5dbe0106e542e24b9ef1b6169e31
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Changes include:
- Neutron external network now configured post deployment
- IP addresses assigned to admin and public OVS bridges if missing
Change-Id: I7a3e8822b43640578db809c3f4d55f399383c1f1
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
Change-Id: I2dc1bf7ffebb6a83face87a62a81cfb15f9a0689
|
|
Change-Id: I482e84bbe3f21f425e6bce74fdfa16f71bf12654
|
|
Tested manually and verified on LF pod1
Change-Id: I882ad9beef003a795585ffb8542aa11aecb67949
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
This applies only to baremetal deployments. Interfaces from admin and
public networks are attached from the host to the OVS bridges brbm and
brbm2 for external connectivity.
Change-Id: If3f5b33eab267e05ee9094396daa40f37e69319a
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Changes include:
- ONOS artifacts are cached on build server until moved to OPNFV artifacts
- tht patch is now unified to single opnfv patch, based on upstream
commit 0f41e0d0309c7bc061dbaed6a8c1e38cb4bdfe19
Change-Id: Iea3d14773dded6c9709c08ef1beb47b254582860
Signed-off-by: bob zhou <bob.zh@huawei.com>
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Change-Id: I4e16b4a4432c935715f0522956614c891a695f70
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
Changes include:
- Fixes Intel Pod 2 inventory file
- Check for DHCP server on the host and disable
- Adds realistic+common network-settings file
- Modifies baremetal deployments to bridge to correct interface
- Adds private/storage network OVS bridges
- Parses network-settings into valid network-environment variables
- If certain network-settings are missing they will be auto-detected
Note: The actual settings set forth for deployment only include
admin/external networks at the moment. Private/storage networks will be
handled in an upcoming patch.
JIRA: APEX-50
Change-Id: I0a1a86f37c08702a93fe167688c3149ba5573db4
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
|
|
Testing if this fixes current CI breakage
Change-Id: Ib4f75a9a1772c5363afab6d24b7a254f8cf6f6f8
Signed-off-by: Tim Rozet <trozet@redhat.com>
|
|
unnessesary"
|
|
If there are dedicated HW nodes for controller and compute nodes,
it should be possible to define node roles in configuration files.
This patch introduces "capabilities" property into
pod_example_settings.yaml so as adds needed flavor definitions
and overcloud deploy parameters into opnfv-deploy script.
Change-Id: Ie62d3fa215aa168390d074c73fdb2fa03ecc67f5
Signed-off-by: Viktor Tikkanen <viktor.tikkanen@nokia.com>
|
|
Change-Id: I4a585a5e72872fa1887aa4487ebd7b17d07feacc
|
|
- enabling subnet configuration for undercloud
- enabling external network configuration
Change-Id: Id1951d31faa1a68f850a75e2466d30b8abcea8ac
|