This reverts commit 1c85fdc390.
Do not use randomly generated strings in configmaps as this leads to
whole helm release redeployment even no values are changed. The random
items have to be generated outside of helm chart and provided via
values.
Also previous behaviour didn't allow to use cache during rolling upgrade
as new pods were spawned with new key.
Change-Id: I423611b18fca0d65e2e721a9c6a0c3d8df0813d2
The gotpl script that determines if a cinder backend is ceph-backed
is not properly handling additional backends that does not have the
driver "cinder.volume.drivers.rbd.RBDDriver". This patch set fixes
the gotpl so it no longer causes a rendering problem.
Change-Id: I902e82301019531832afebce7a1e2f0b28bac8f3
Signed-off-by: Tin Lam <tin@irrational.io>
This change refactors the apparmor job to utilize the feature
gates system instead of relying on separate scripts.
Also disabled barbican running in the apparmor job temporarily
until the correct profile gets used and it can deploy
succesfully.
Change-Id: Iadacd214de3fdb06e4acde4433c5fa86973371d5
This patch set fixes an issue with where the keystone chart's
domain-manage job/pod always restart once due to a calculation
logic error.
Change-Id: I801d04559a526d3a7339cd5102f2e738af9f72e0
Signed-off-by: Tin Lam <tin@irrational.io>
This patch set adds in a script that cleans up orphaned or
lingering rally helm test pods.
Depends-On: https://review.opendev.org/#/c/683759/
Change-Id: I94fc8d067b421248cf74fe40b2e8520f63d4417c
Signed-off-by: Tin Lam <tin@irrational.io>
It was observed that when increasing amount of
conductor workers from default "1" to higher value
the readiness probe fails to check rabbitmq connections
for conductor processes - it happens since the script is trying
to obtain rabbitmq connections for parent conductor process
which in case of workers>1 doesn`t open rabbit connections
but spawns child processes which handle rabbitmq
connections instead.
This patch removes the "check-all-pids" option, keeps the logic
but simplifies and fastens he code - instead of checking all
processes when "check-all-pids" option was set (however
regardless of "sock_count value" if only one process opens connection
the check returns positive result) processes will be checked one-by-one
until the first one with open rabbitmq connection(s) is
found.
Change-Id: I72be0bbdefcba77a55b6ceed6e192c9621c069eb
The 'options' keyword for setting mtu in 'set interface' does not
set mtu and it seems to ignore/fail the request silently.
Change-Id: Icec98c5166611a8c538f93e6326cf7d20b545ecd
This patch set addresses a failure in the compute-kit network
policy failing as some application:nova to application:nova
pods communication is blocked.
Change-Id: I29cc044e0d4f10198c23c7c3e132ab0093f91e21
Signed-off-by: Tin Lam <tin@irrational.io>
When the default release was switched from ocata to stein, some of the
policies were duplicated. This moves the ocata overrides back to where
they belong, and adds overrides for pike, queens, and rocky.
Change-Id: I342d69e721b2692987951055e41ed5e153a91d6c
This patch set updates the default job to use OpenStack Stein release.
The previously default Ocata release will be place in separate job.
Change-Id: I489324f762a179a2cab5499a6d8e57e97c81297f
Signed-off-by: Tin Lam <tin@irrational.io>
This PS updates the gateway setup script to support deployments where
multiple default routes may be present, eg when connecting to a vpn.
Change-Id: I5fe86e9471ecb6a9c80e1e4f942822cac00da669
Signed-off-by: Pete Birley <pete@port.direct>
This patch set adds in a capability for the user to defaultly use a
FQDN for the nova compute hostname and the hypervisor hostname when
the host is not explicitly specified in the .Values.conf override.
Change-Id: I3243068dfe91ebb97b3885002296a0f454822ec5
Co-authored-by: Drew Walters <andrew.walters@att.com>
Signed-off-by: Tin Lam <tin@irrational.io>
Because it's almost time for expiring on some python version, OpenStack client
running on that version generates some messages for warning. Two scripts on
nova Fixed by this PS get version information using the OpenStack client
without any protection for this kinds of messages. This PS gives a little
more sophisticated way of it.
Change-Id: I2896c76e012b9acbf1e725276ba9c0b74789fa54
With the update of openstack clients:
openstack client >= 4.0.0
neutron client >= 6.14.0
neturon lib >= 1.29.1
The command 'openstack network show ${network} -f value -c subnets'
returns '[]' instead of null string if no subnets found in the
specific network. This commit adds a check logic to avoid subsequent
command returns error by using '[]' as subnet input.
Change-Id: I7e7d5209227b0e34131b7715dbd3faa6066a94b7
Signed-off-by: Mingyuan Qi <mingyuan.qi@intel.com>
This patch set specifies the barbican egress in the override to be
inline with other services.
Change-Id: Ie4997a5bb476a21253e98e4c69a77e58b8498f4f
Signed-off-by: Tin Lam <tin@irrational.io>
Signed-off-by: Tin Lam <tlam@omegaprime.dev>
This disables the keystone-auth single node job and all multinode
periodic and experimental jobs while standing issues with the
kubeadm-aio image deployment are sorted out
Change-Id: I4e1de001ddf17b3c035ca174b7ef8acec8f2bf2c
Signed-off-by: Steve Wilkerson <sw5822@att.com>
- Change all tests to support Nautilus,Mimic and Luminous releases
- Update ceph-config-helper image
Change-Id: I557b1efa12529d0ee51d4c5b9d4beb4abf1b0574
Currently using envsubst to perform substitution of value overrides in
the feature gate caused conflicts as gotpl gets templated into those
overrides. This adds in '%%%REPLACE_${var}%%%' and uses sed to perform
the substitution instead to address the issue.
This is to achieve parity with OSH-infra patch in [0].
[0] https://review.opendev.org/#/c/697749/
Depends-On: https://review.opendev.org/#/c/697749
Change-Id: I3ed504c65900e7b84728019f3acdf706a40c0427
Signed-off-by: Tin Lam <tlam@omegaprime.dev>
This patchset adds the capability to the Nova chart to be able to wait
for a percentage of the compute nodes/hypervisors to become ready/available
before continuing on with the deployment. It will be disabled by default,
because this is a feature that may or may not be needed in production
deployments.
Change-Id: I971151a663afc87e7d62efa4ab3723c5472a3736
This ps update neutron ovs agent to support properly:
- setting mtu on dpdk bond and nic interface port
- setting vhost-iommu-support on dpdk bond and nic interface port
- setting n_txq values on dpdk bond and nic interface port
Change-Id: I422fa21a622642ecb7c49914fef04073e4f984bc
This change adds in information about overriding the dns entries
that can end up overwriting the existing ones in resolv.conf
and resulting in a lack of network connectivity when deploying
AIO behind a proxy.
Change-Id: I9faab36a114e28fe8f4c312e7cd64a582333033b
