It's possible for Prometheus to return multiple metrics, where
each of them has different labels. Before this the client
would fail with error when using the default table formatter,
because of for example:
Row has incorrect number of values, (actual) 8!=9 (expected)
It was also possible to mismatch the label keys and values
for some of the rows.
This patch makes sure the label values are matched to the correct
keys. It also makes sure it uses all labels and it fills
missing values with empty strings for metrics, which don't have
that label defined.
Example table output:
+------------+----------+------------+----------------+-------+
| group | __name__ | job | instance | value |
+------------+----------+------------+----------------+-------+
| | up | sg-core | localhost:3000 | 1 |
| production | up | prometheus | localhost:9090 | 1 |
+------------+----------+------------+----------------+-------+
Example json output:
[
{
"__name__": "up",
"group": "",
"job": "sg-core",
"instance": "localhost:3000",
"value": "1"
},
{
"__name__": "up",
"group": "production",
"job": "prometheus",
"instance": "localhost:9090",
"value": "1"
}
]
Change-Id: Id0dfabf52fe0a21194c498f4aefa1bff1d3eeea9
This adds a new config option: "prometheus_ca_cert". If this
option is set, then it forces the client to use https to access
prometheus and it uses the specified ca cert to verify the
prometheus' certificate.
Change-Id: Iccb911a590d5b3b9a4c6ac08c4d020641c8094a9
This adds new functional tests, which are supposed to be run
on devstack with a running instance of prometheus.
It tests all of the cli commands as well as all the functions
exposed in the python client.
These tests could be included into the telemetry-dsvm-integration
jobs in the future to use the same devstack vm.
Change-Id: Ibd6deec559465bf3cb7480681b816f55bdf9010e
Something must have changed either with cliff or
with a different version of python. All of the
cli commands end with: "'Namespace' object is not subscriptable"
This is caused because of how the parsed_args fields
were accessed.
This patch addresses the issue. Now it accesses the fields
the same way as aodhclient does. The unit tests for cli
now use the arg parser to actually test this.
During this I also discovered incorrect mocking in the
configuration tests. I added another mock to the test cases
which were missing it.
Change-Id: Ib5dbbdb48bdfd3214ec76acc4c68649e25f695db
Adding comments for all ignores and extensions.
Adding extensions from OpenStack Hacking.
Fixing few breakages of those added extensions.
Change-Id: Idacee2ca555411e33b65817fb9245d130cf36574
This change fixes python 3.8 unit tests.
Unfortunately using grouping parentheses in with statements
is a python 3.9+ feature. I had to replace the parentheses
with a less elegant "\" to escape continuation lines.
I also included a .zuul.yaml, without which this couldn't
get merged and a .gitreview for convinience.
The telemetry-dsvm-* tests are non-voting for now.
There isn't any relevant test for this repositary
there as of right now,
those will get added in the next few weeks. And
unfortunately all the telemetry-dsvm-* tests fail
due to pyparsing version mismatch right now.
Once I or anyone else adds relevant tests to the
telemetry tempest plugin, we can make the tests
voting here.
Co-authored-by: Martin Magr <mmagr@redhat.com>
Co-authored-by: Erno Kuvaja <jokke@usr.fi>
Change-Id: Icc7b0229bca0664ee7fd60e3932df8f599beb500
* Add unit testing
* Fix code according to CI
This includes:
- formating changes
- rewording of some doc strings
- adding support to {label!~'value'} in rbac
* Add unit tests automation
* Fix CI automation
* Add requirements.txt
* Remove old observability client
* Add initial functionality for prometheus querying
* Fix a copy-paste error in get_client()
* Add additional functionality.
This commit adds:
- commands:
delete
clear-tombstones
snapshot
- Better rbac injection as well as a possibility
to disable rbac.
- Configuration of prometheus_client through
env variables and /etc/openstack/prometheus.yaml
* Make README up to date
* Implement Martin's PR comments
* Implement better support for label values in rbac
* PEP8
For deploy.discovery command:
This patch allows inventory file destination fallback override and adds
additional fallback paths used by standalone deployment.
For deploy.setup command:
This patch allows to override usage of generated inventory file.
This patch adds basic functionality of the plugin. It successfully registers as openstackclient plugin and contains two basic observability commands:
- discover
- prepares ansible inventory file with overcloud and undercloud nodes and gather data for prometheus agent according to which nodes are scrapable
- setup
- starts proper ansible playbook based on component (currently only prometheus_agent is available)
Co-authored-by: Marihan Girgis mgirgisf@redhat.com
Partially-Implements: OSP-14664
Related: infrawatch/osp-observability-ansible#11
