Devref contains a document discussing project priorities. Include a
link to the list of specific Kilo priorities that lives over in the
nova-specs repo.
Change-Id: I1770447b03df1cc1ab029780fada96fab8ccaa0f
The unit testing devref had a bit of inconsistent formatting, and a reference
to HACKING.rst which could have been linkifed. This patch fixes both of those,
and additionally the examples have been updated to reflect the recent split
into functional and unit tests.
Change-Id: I15129808d0d1271c1223e7ab8d8532f8cb2f7293
The internationalization devref was out of date (and wrongly named). This
patch documents the current advice for internationalization.
Change-Id: If91b3b11c4050cb231b7d9c38b6beca8ba6ce236
Replace URLs for workflow documentation to appropriate parts of the
OpenStack Project Infrastructure Manual.
Change-Id: Ice00abcbb431b29e28beb0cd8810b8653ef66139
The diagram on http://docs.openstack.org/developer/nova/devref/vmstates.html
does not currently show the "SHELVED_OFFLOADED" state and the
associated "shelve", "shelveOffload" and "unshelve" state transitions.
Change-Id: I50eb07a84336160b8120ca294598576eeb982d28
Closes-Bug: #1394215
This patch makes the NUMATopologyFilter and instance claims on the
compute host use instance fitting logic to allow for actually packing
instances onto NUMA capable hosts.
This also means that the NUMA placement that is calculated during a
successfull claim will need to be updated in the database to reflect the
host NUMA cell ids the instance cells will be pinned to.
Using fit_instance_to_host() to decide weather an instance can land
on a host makes the NUMATopologyFilter code cleaner as it now fully
re-uses all the logic in VirtNUMAHostTopology and
VirtNUMATopologyCellUsage classes.
Change-Id: Ieabafea73b4d566f4194ca60be38b6415d8a8f3d
Closes-bug: #1386236
This covers the plan for zero downtime upgrades. It looks at all the
building blocks we have in place, and a few we have planned for kilo and
pulls that into a process that we hope will deliver zero downtime
upgrades.
This is just an initial stake in the ground to help drive the upgrade
efforts during kilo.
Change-Id: I3974d18f79ba655257dcac49208821bc191d8cc5
The rpc doumentation under nova developer reference contained typos. These typos have been fixed
Closes-Bug: #1393135
Change-Id: Ib8e76625b874737489c7618102309da16835deba
Remove affinity filters that were previously deprecated. Filters
originally deprecated in 2bca0c9011
Change-Id: I5f642988dbd35e02e13dd13c69c6625cf9b07237
Now that we have stopped running python26 jobs in the gate, remove
python26 from the list of jobs in tox
project-config patch: Ib495467e00e4b6abf2b3c2b9445b51652ba25167
Change-Id: I18262e91c9f5fc2381c6f5b37eb45b5f27aa81db
Add a new nova scheduler weighter, sort the filter
hosts according to host io ops number, aims to
booting instances on light workload hosts.
DocImpact: Adds io_ops_weight_multiplier to [DEFAULT]
group of nova.conf and an new default schedule weigher.
Change-Id: Ib3c9184f10b2ebe6b1230365a51b5542dffd447c
Implements: blueprint io-ops-weight
This patch removes the baremetal virt driver from nova, which was
deprecated in Juno. It does not aim to perform any of the refactoring
we need to do now that the driver is gone, but is intended as a clean
removal of all the baremetal code itself.
Of specific note are the api_samples tests. We didn't have samples to
test the ironic proxy code, so I removed the baremetal ones for show
and index, resetting that extension back to "untested" status as if it
was new.
Tests that stub out ironic interactions and re-introduce those tests
should be added after this patch.
Change-Id: Ia76e41a8a3b7230701872ae7a1975edc3d9ea847
The nova-manage flavor sub-command was deprecated in Juno milestone-1
via commit a601556684 with a note that it
will be removed in Kilo.
This removes it.
Change I4e2399f2fbab2c20299dd62c104b9220c1d36fb2 is to remove the
related Tempest tests.
Change-Id: I1798c8a0697009b3dcebeb818580c8323880e395
Document what APIs nova considers to be public contracts, anything not
in this list should be considered a private API.
Clarify the 'API' section to make sure its clear we are talking about
the REST API.
Change-Id: If71adc7a9d5dc1bde79f75d10e67500f6f92b81f
There is an option "disk_allocation_ratio" in DiskFilter instead of
a typo "disk_allocation_ration".
This patch fixes the typo on filter_scheduler.rst.
Change-Id: Iad4951fbdca97eccb956c42270bfc170fc7c7dac
The syntax suggested for adding hooks to nova via setup.py entrypoints is:
entry_points = {
'nova.hooks': [
'resize_hook': your_package.hooks.YourHookClass,
]
},
But this is incorrect. The proper syntax for an entry_points entry is:
entry_points = {
'nova.hooks': [
'resize_hook=your_package.hooks:YourHookClass',
]
},
Change-Id: Id7bc6891dc542d4317e82e15300c7293eb0d17a0
Closes-bug: #1374666
There were broken links on the following class
'AggregateNumInstancesFilter', 'RamWeigher' in
document filter_scheduler.html. Fix these broken
links.
Change-Id: I3cac48a6ad466a32a88c23c9f128e66dae5889ee
Closes-Bug: #1363955
Don't list entire api/autoindex under Developer Docs, we already have a
model index link at the bottom of the page, showing the entire
autoindex on the home page is makes the page overwhelming.
Change-Id: I25c5b50412881b55aa745ce3da2c22466e879fcf
At the nova mid-cycle we agreed that while specs have made our blueprint
process significantly better, we think the current approach still leaves
much to be desired. This document is the outcome of the discussion at
the mid-cycle on how we want to manage blueprints in Kilo.
This proposal is being placed in devref, as we want to use gerrit's
workflow (two core reviews, inline comments, revisions etc) and this is
a document for developers.
Introduce the idea of project priorities.
Change-Id: I7075a130c01be22f3ee5315d4a65b8e524b5a1ad
* Devref mentioned 10.10 instead of 14.04 as the tested ubuntu
environment
* Point to our official git mirror instead of github
Change-Id: I4514faeaebe200a2f616bc37fca02c2927edf599
This patch adds the basic NUMA filter that will take the proposed
instance topology and try to match it against that of a host.
The matching will be done in the following manner:
* Filter will try to match the exact NUMA cells of the instance to those
of the host. It *will not* attempt to pack the instance onto the host.
* It will consider the standard over-subscription limits for each cell,
and provide limits to the compute host accordingly.
* If instance has no topology defined, it will be considered for any host.
* If instance has a topology defined, it will be considered only for NUMA
capable hosts.
DocImpact
Change-Id: I8788dde69524c8a32a41ce31a96c89f9b09e91ce
Blueprint: virt-driver-numa-placement
At the nova mid-cycle we agreed that while specs have made our
process significantly better, we think the current approach still leaves
much to be desired. This document is the outcome of the discussion at
the mid-cycle on how we want to manage blueprints in kilo.
This proposal is being placed in devref, as we want to use gerrit's
workflow (two core reviews, inline comments, revisions etc) and this is
a document for developers.
It turns out adding the specs requirement to blueprints adds a
non-trivial amount of overhead, revisit when a blueprint and spec is
needed to minimize overhead in cases where it isn't helpful.
Change-Id: Iaaadc26f3f51ee091419beaa12e0508623eed99e
Remove intersphinx from the docs build as it triggers network calls that
occasionally fail, and we don't really use intersphinx (links other
sphinx documents out on the internet)
This also removes the requirement for internet access during docs build.
This causes docs jobs to fail because we error out on warnings.
Change-Id: I71e941e2a639641a662a163c682eb86d51de42fb
Related-Bug: #1368910
While setting up a new Ubuntu Trusty development
system without devstack I ran into an undocumented
native package dependency for sqlite, so add those
here.
Change-Id: Ib82ea58e85f6ccab6008f2d19a31d0e890136ca7
For some time, we've been informally implementing a "revert early"
policy around handling -2 worthy complaints after a patch has been
merged. These reverts can be contentious, so it's appropriate we make
it very clear what the policy and procedure here is in order to lessen
the contentiousness of these reverts.
Co-authored-by: Michael Still <mikal@stillhq.com>
Change-Id: I90e4153a80c859a928630809aef31e7313f08906
Adds a websocket proxy that is compatible with Nova
serial ports.
Co-Authored-By: Vladan Popovic <vpopovic@redhat.com>
Co-Authored-By: Ian Wells <iawells@cisco.com>
Co-Authored-By: Sushma Korati <sushma_korati@persistent.co.in>
Partial-Implements: blueprint serial-ports
Change-Id: Ia944cb93945140e6341588063329a981f7e778f1
Adds a filter AggregateIoOpsFilter which provides the ability
to read from aggregates metadata "max_instances_per_host".
DocImpact
Implements: blueprint per-aggregate-max-instances-per-host
Change-Id: I569aab7842d2c67042b89c75941a4664a599a7f6
Adds a filter AggregateDiskFilter which provides the ability to
read from aggregates metadata the "disk_allocation_ratio".
DocImpact
Implements: blueprint per-aggregate-disk-allocation-ratio
Change-Id: I79d59fcdfb09e67ed6f12113615c673624b24a19
Jenkins