delorean-repo:
replaced by tripleo-repos and other repo setup approaches
hosts:
heat hasn't manage hosts file for many releases
localboot:
replaced by diskimage-builder grub2, bootloader elements
ntp:
replaced by installing chrony package
os-net-config:
the package is installed by other elements, and the script is now
called from ansible
overcloud-dracut:
replaced by diskimage-builder dracut-regenerate element
overcloud-partition:
no image now uses this, overcloud-hardened-uefi-full uses
overcloud-partition-uefi
Change-Id: I990493b51d87a40140c1648f2b04c982ce959b1d
- Ceiling of pbr version can create installation conflicts with pip.
- Fixes tox default env list to be python version agnostic
- Updated template to use newer jobs (include testing with newer python)
Bug: #1921679
Change-Id: Ifd0d2f810799992faa1db41b80bf93bc97d5d19d
With RHEL8, net.ifnames is set to 0 in the grub config. This is a change
in the default image which can have user impacts. We want to clear it to
go back to the previously defined behavior in RHEL7.
Change-Id: I621066ea5e0e934668ff4c91e991a95d18e019f7
Closes-Bug: #1841441
Add boundaries to the matching to ensure that we don't incompletely
match the IP.
Closes-Bug: 1590010
Change-Id: I06c1e145fd5e98827d8b092719f11c642b1ecdf1
This patch corrects the Heat metadata key we look for when trying
to determine the Swift metadata URL. Previously we were using
'os-collect-config.swift.metadata_put_url' which is incorrect,
thus causing us to skip the metadata_ip check. The correct
value to look for when using Swift temp URL's
is 'os-collect-config.request.metadata_url'.
This fixes an issue that broke network isolation upstream due to
the fact that the metadata_ip check was getting skipped when using
Swift temp URLs for signal reporting. The metadata_ip is required
to when using network isolation because it gives time for the
network connection to become active before proceeding. Not allowing
the network connections to properly activate can cause intermittent
and or all notifications to Heat to fail.
In the cases I saw locally this happened when using the t-h-t
single-nic-with-vlans network isolation templates on the compute node.
Change-Id: I7c17e3feec790f1e7760f33939c7378bb070fbc0
Closes-bug: #1550380
This patch removes /etc/udev/rules.d/99-dhcp-all-interfaces.rules
once os-net-config has executed at least once successfully. This
will ensure no more DHCP bootstrapping occurs until we get more
information on why dhcp-all-interfaces.sh is leaking DHCP processes.
Change-Id: I5b456419df6c6606a4bda1202ea49378e6feb6b2
Related-bug: #1538259
This patch adds a new configure_safe_defaults function that
is called on EXIT. If the exit code is non-zero the function
executes code to configure DHCP on all active interfaces.
Since DHCP is used to initially bootstrap all TripleO instances
via the provisioning network this should always be a safe default
and allow communication with Heat to persist if an invalid
(breaking) network configuration change is pushed out. This would
facilitate immediate Heat notification of the failure and subsequent
Heat updates to potentially help correct the invalid network
configuration.
Change-Id: Ibccba1ee77aa13c85a78a75d2e6cd179f87d7d16
Depends-on: Ibe0e32bc09979bc68b92a722b2bfa383e77502a9
Updates os-net-config to use --detailed-exit-codes so that when
a network change has been detected we can perform a network validation
via a new ping_metadata_ip function. This function performs a simple ping
test back to the IP address of the Heat API which should always be
available after any network configuration change.
The motivation for this change is to allow network configuration
changes time to settle. When making changes to the ctlplane (which
is typically used to contact the undercloud Heat API for subsequent
configuration notifications and updates) it is sometimes desirable to
wait a few seconds before proceeding. This avoids a race where the
ctlplane was switch to a static IP address and the very next
Heat notification would sometimes fail.
Depends-On: I8f22fa15335d1276f4e444a6454a24ff486e1495
Change-Id: I257e1cba6dee16f73f75512d1284e1e3b9d4c831
The version of pbr we were using was having trouble parsing a new version
identifier for python-qpid, We have the pbr version hardcoded in many
places so update it to match whats currently in global-requirements.
Fixes bug #1470871
Change-Id: Ia5aec472555ad1368684b224a55c29242ae40b58
We standardized the logging options in os-net-config in
I31005e4e10372018685a93dc2f80619f16c513d9. Previously
os-net-config was functionally logging at the info level
however, which is what we want.
This patch switches all the os-net-config commands to use
-v INFO level logging again.
Change-Id: Ia90a83da27cd4767425c4913fd150d5df5c3163e
This makes sense because pretty much all our low level
networking models make use of openvswitch. Also,
you need Openvswitch to be installed (and running)
before any of the os-refresh-config hooks for
os-net-config will work.
Change-Id: Ieff88a48a1e28fa32c343ae1eeea9fec4aa3df26
We were installing some packages with support for prerelease versions in
expectation of needing it in our CI. This turns out to not be needed so
we should remove it.
Change-Id: I7ae440d3534e33144942c2230a88cf54f8a153d3