With this patch, ceilometer-polling is able to expose
metrics itself in Prometheus format so they can be
scraped directly from a Prometheus instance.
Change-Id: I68d1dc96a89e57cafa9dba207d92403a1e156fe6
Add file to the reno documentation build to show release notes for
stable/2024.2.
Use pbr instruction to increment the minor version number
automatically so that master versions are higher than the versions on
stable/2024.2.
Sem-Ver: feature
Change-Id: Ia23342dee322e2da9373aa1c77e05c6d3655e35d
This patch adds possibility to start child service for heart beat reports
of polling agents. Such data can be leveraged in service health check
scripts for monitoring purposes.
Change-Id: I721a1ef997c5dfb59cdd0e759fcd7d511c24bbb0
When polling libvirt for instance data, we weren't getting
the instance metadata. In particular we weren't getting the
metering.server_group, which is useful for autoscaling.
The metadata need to be retreived by an API call to nova.
The code already called nova API to get flavor ID. I adjusted
the code to get the flavor ID and instance metadata by the same
call, so the number of API calls is unchanged.
Change-Id: I8034fca8a86263f38e396c4c1512546863a754db
This has been deprecated for removal [1]. Use the stdlib variant
instead.
[1] https://github.com/testing-cabal/testtools/commit/59b890db3c
Change-Id: If4ed661efa40dd5aed8c1784c1023d01dd81b23a
Signed-off-by: Stephen Finucane <stephenfin@redhat.com>
It was deprecated during the previous cycle because of retired
WinStackers project[1].
The following pollsters are also removed because these are supported
only by Hyper-V inspector.
- disk.device.latency
- disk.device.iops
[1] 7660339b4a4d578f2615fd6c27e8c9627ab29d37
Change-Id: I872f93b945233a1288c773302be3835db2fd2e4f
When events are received from cinder, this includes the the volume_type_id
field (and not the volume_type name). Since Wallaby version it is now
possible to fetch this id from the API, so we can now make this poller
consistent with what is sent over the notification bus.
Another useful missing field for backups is the is_incremental field.
As the backup.user_id field was added in microversion 3.56, it makes sense
to add this information here as well to make it consistent with other
samples.
Closes-bug: #2064290
Change-Id: Ib109f399bab43ed8a6a0e0afea27ff0e5f7ae883
This previously raised an KeyError, as the 'id' was pop'ed inside the
try-except statement, so when the exception was catched the key has already
been removed.
Closes-bug: #2064124
Change-Id: I617d9149d1ff45cb81b71feeb3097ac1f9cd0116
The following metrics are all specific to OpenDaylight but
OpenDaylight support was already removed and we no longer need to
maintain these.
- switch.flow
- switch.table
- switch.port.*.drops
- switch.port.*.errors
- switch.port.receive.*_error
- switch.port.collision.count
Change-Id: I2f5f71d0a1ed7de0e0886e104445b087bdc05d67
These are detected as errors since the clean up was done[1] in
the requirements repository. Minimum versions are bumped to avoid
installing these known bad versions.
[1] 314734e938f107cbd5ebcc7af4d9167c11347406
Also remove the note about old pip's behavior because resolver in
recent pip does not require specific order.
Change-Id: I04335dd0d06b5c65b805fbb1e0fc0cd1c10a79fb
The feature was deprecated during the previous cycle[1], because
the netwokring-odl plugin[2] was already retired.
[1] a92d77b1823f7d9789d07faca785299637cabaa6
[2] https://opendev.org/openstack/networking-odl
Change-Id: I3437b73dea37f600a505f908c14e54536a40f588
If the credentials are not set correctly, the objectstore
poller will throw a stack trace trying to access swift
tenants where response is 403 Forbidden. We already catch
and handle gracefully the 404 in the endpoint code,
extending that to cover 403 with meaningful error message
logged instead of raising the error.
Closes-bug: #2061759
Change-Id: Idfa050f7c97941cbe1a4f40b98e0264bce548164