When selecting a single instance, the query passed to gnocchi seems to be incorrect.

params: {filter: "instance_id in cb5f8625-155c-4528-a51e-
"Invalid input: expected a list for dictionary value @ data['in']['instance_id']”

We're using this dashboard: https://grafana.com/dashboards/4569. The plugin could probably ensure that this is actually handled as a list.

When using token auth mode and server access mode, test operation will get gnocchi resources through grafana server normally.
However, if using keystone auth mode, test operation will get token through proxy but get gnocchi resource directly.
For example (in one test operation):
http://<grafana_ip>:<grafana_port>/api/datasources/proxy/5/v3/auth/tokens
http://<gnocchi_ip>:8041/v1/resource

I have built a grafana server and could connected gnocchi datasource through Http. But after our OpenStack env. (include gnocchi) swift from Http to Https, it can not connected anymore.

Does grafana-gnocchi-datasource support Https ?

Right now it's impossible to sort all obtained measures or show only part of them.

For example we need to get instances with top CPU utilization. That means that we want to get all metrics (cpu_util) for some resource type (instance), sort by value in descending order and show only some of them in Grafana.

We can use "resource search" query type, set resource type, metric name and some simple query string (say, started_at>='somedate'). After that we will receive some amount of metrics, maximum amount of those metrics will be equal to api.max_limit value in gnocchi.conf.
If we want to get all measures (to perform some kind of analysis) we have to set big value for api.max_limit option in gnocchi.conf. But in this case we will get thousands of metrics on one chart in Grafana.

What we need is some kind of functions to set them on all of our received values.
There are some example implementations:
Graphite datasource: http://docs.grafana.org/features/datasources/graphite/#functions
Zabbix datasource: http://docs.grafana-zabbix.org/reference/functions/

Using Auth Mode keystone, with all settings OK (URL endpoint http://controller:5000, domain, project, user and password) and CORS enabled.
Grafana gets the Keystone token and the Gnocchi endpoint, and goes to query it.
But, the connection is done via HTTPS automatically forced, not HTTP.
As HTTPS is not responding, data load fails.
This happens in some versions of Firefox and Chrome, Internet Explorer is not having this issue and gets the data OK.

When there are missing points from gnocchi, e.g. there are no data with the granularity in the metric, the missing values are represented as 0 value. This makes the graphs awkward, e.g. when ceilometer samples the CPU usage in 10 minutes interval, but using the 5 minutes aggregation in Gnocchi, the graph looks like a sawtooth wave. I think missing values should be null and the null values can be interpolated then.

Hello,

i recently deployed grafana with gnocchi in an openstack kolla-ansible deployment.
Now I have the Problem that I am unable to query for instancesnames like it is shown in your documentation.
When I query for

resources(instance, id, host=COMP2)

i get a popup with:

Templating
Template variables could not be initialized: Unexpected token h in JSON at position 0

In the gnocchicli it looks like this:

gnocchi resource list --type instance -f json [ { "started_at": "2018-01-16T12:54:19.093064+00:00", "user_id": "399af2114fb043aab626b7c022e5cfaa", "revision_end": null, "creator": "6509a0fb8ecf40cf98b77ad311bad32d:c61f928dd839427bbde384ac8151ad8b", "revision_start": "2018-01-16T12:54:19.093084+00:00", "flavor_name": "m2.medium", "host": "COMP2", "image_ref": null, "flavor_id": "f8464ff2-f60d-45b9-8198-973cd0e11acb", "server_group": null, "original_resource_id": "08176e67-81ad-43ce-8a72-847849dfbbfc", "ended_at": null, "project_id": "0fbb5957dd564cb1928e2a64888c096e", "type": "instance", "id": "08176e67-81ad-43ce-8a72-847849dfbbfc", "display_name": "test" } ]

Hello,
we are using the current stable Gnocchi Grafana Datasource plugin in to query the metrics of an OpenStack cluster. At the moment, I am trying to use the dynamic aggregates query type to fill a Single Stats panel. However, the return value of the query is an array of items instead of a single value. What am I doing wrong?

The expected output of the query is the sum of the mean number of vcpus of all instances for a given search string in the given time period (I know I can do that via the resource search (aggregated measurements) but would like to apply some math to the result).

The operation I have defined is: (aggregate sum (metric vcpus mean))
and the response I can see in the Query Inspector looks like this:

...
response:Object
measures:Object
aggregated:Array[4]
0:Array[2018-03-13T13:25:00+00:00,300,16]
1:Array[2018-03-13T13:50:00+00:00,300,192]
2:Array[2018-03-13T13:55:00+00:00,300,48]
3:Array[2018-03-13T14:00:00+00:00,300,224]
references:Array[32]
0:Object
...

How do I aggregate the response array into a single value?

In order to produce graphs for things like "total allocated storage by project", it would be great to have support for the "groupby" operation in the gnocchi data source (for both dynamic aggregates and the older aggregate api).

Some response jsons will include “Infinity”, which JSON.parse can’t handle.

E.g. "{"measures": {"aggregated": [["2018-10-10T13:10:00+00:00", 300.0, Infinity]]}}"

The plugin could probably replace those occurrences with 1e+99999, avoiding such parsing errors.

Hi,

The Gnocchi Grafana datasource is failing with Gnocchi is unreachable or CORS is misconfigured (detail: TypeError: NetworkError when attempting to fetch resource.) Grafana version is 8.2.1 and gnocchixyz-gnocchi-datasource has this fix applied.

Both Keystone and Gnocchi have this in their configs:

[cors]
allowed_origin = *
allow_methods = GET,POST,PUT,DELETE,OPTIONS,HEAD
allow_headers = X-Auth-Token,X-Openstack-Request-Id,X-Subject-Token,X-Project-Id,X-Project-Name,X-Project-Domain-Id,X-Project-Domain-Name,X-Domain-Id,X-Domain-Name

But somehow the integration keep failing.

I am using gnocchi version 7.0.1 with OpenStack Queens.
Can anyone send any sample query to get the information from grafana with gnocchi plugin server?
I mean what are the information, I have to write for the following fields:
Resource Type: instance
Query: What should I write here?
Operations: What should I write here?

Kindly write a sample query to get the cpu utilization of an instance.

Hi,

I'm trying to connect Grafana to Gnocchi data source with keystone authentication.
But I don't have permissions to modify the gnocchi.conf or keystone.conf for [cors] "allowed_origin".
Is there any way where I can access Gnocchi resources? e.g. Passing the required headers in HTTP request itself? Or extending the gnocchi.conf as we do it for apache conf.

when enabled multi value feature in template and configure the panel repeat for that variable, the panel graph show nothing.

When trying to add a gnocchi datasource i receive
Keystone failure: (404 Not Found)
And in the console of the browser i can see the following error
POST http://192.168.37.250:3000/api/datasources/proxy/8/v3/auth/tokens 404 (Not Found)
When looking for this file i can see that there is a file under v1 but not under v3

@sileht @jd I've read all docs, but couldn't I fix the issue.

Error :

Error

Gnocchi error: No response status code, is CORS correctly configured ?

Config files :

Gnocchi.conf ↴

[DEFAULT]
debug = true
verbose = true

[api]
workers = 4
middlewares = oslo_middleware.cors.CORS                   
middlewares = keystonemiddleware.auth_token.AuthProtocol  
paste_config = /usr/lib/python2.7/site-packages/gnocchi/rest/api-paste.ini
auth_mode = keystone


[archive_policy]

[cors]
allowed_origin = http://172.18.1.199:3000


Auth-Token,X-Subject-Token
allow_methods = GET,POST,PUT,DELETE,OPTIONS,HEAD
allow_headers = Content-Type,Cache-Control,Content-Language,Expires,Last-Modified,Pragma,X-Auth-Token,X-Subject-Token
expose_headers = Content-Type,Cache-Control,Content-Language,Expires,Last-Modified,Pragma

[cors.subdomain]

[database]
backend = sqlalchemy

[healthcheck]
[incoming]

[indexer]
url = "postgresql://gnocchi:[email protected]/gnocchi"

[metricd]
workers = 8

[oslo_middleware]
[oslo_policy]

[statsd]
host = 0.0.0.0
port = 8125

[storage]
coordination_url = file:///var/lib/gnocchi/locks
driver = file
file_basepath = /var/lib/gnocchi

[keystone_authtoken]
auth_type = password
auth_url = http://127.0.0.1:5000/v3
memcached_servers = localhost:11211
project_domain_name = default
user_domain_name = default
project_name = service
username = admin
password = ***************************************
region_name = Region_LAB

Keystone-paste.ini ↴

[filter:debug]
use = egg:oslo.middleware#debug

[filter:request_id]
use = egg:oslo.middleware#request_id

[filter:build_auth_context]
use = egg:keystone#build_auth_context

[filter:token_auth]
use = egg:keystone#token_auth

[filter:admin_token_auth]
use = egg:keystone#admin_token_auth

[filter:json_body]
use = egg:keystone#json_body

[filter:cors]
allowed_origin = http://172.18.1.199:3000
paste.filter_factory = oslo_middleware.cors:CORS.factory
oslo_config_project = keystone


[filter:http_proxy_to_wsgi]
use = egg:oslo.middleware#http_proxy_to_wsgi

[filter:healthcheck]
use = egg:oslo.middleware#healthcheck

[filter:ec2_extension]
use = egg:keystone#ec2_extension

[filter:ec2_extension_v3]
use = egg:keystone#ec2_extension_v3

[filter:s3_extension]
use = egg:keystone#s3_extension

[filter:url_normalize]
use = egg:keystone#url_normalize

[filter:sizelimit]
use = egg:oslo.middleware#sizelimit

[filter:osprofiler]
use = egg:osprofiler#osprofiler

[app:public_service]
use = egg:keystone#public_service

[app:service_v3]
use = egg:keystone#service_v3

[app:admin_service]
use = egg:keystone#admin_service

[pipeline:public_api]
pipeline = cors healthcheck sizelimit http_proxy_to_wsgi osprofiler url_normalize request_id admin_token_auth build_auth_context token_auth json_body ec2_extension public_service

[pipeline:admin_api]
pipeline = cors healthcheck sizelimit http_proxy_to_wsgi osprofiler url_normalize request_id admin_token_auth build_auth_context token_auth json_body ec2_extension s3_extension admin_service

[pipeline:api_v3]
pipeline = cors healthcheck sizelimit http_proxy_to_wsgi osprofiler url_normalize request_id admin_token_auth build_auth_context token_auth json_body ec2_extension_v3 s3_extension service_v3

[app:public_version_service]
use = egg:keystone#public_version_service

[app:admin_version_service]
use = egg:keystone#admin_version_service

[pipeline:public_version_api]
pipeline = cors healthcheck sizelimit osprofiler url_normalize public_version_service

[pipeline:admin_version_api]
pipeline = cors healthcheck sizelimit osprofiler url_normalize admin_version_service

[composite:main]
use = egg:Paste#urlmap
/v2.0 = public_api
/v3 = api_v3
/ = public_version_api

[composite:admin]
use = egg:Paste#urlmap
/v2.0 = admin_api
/v3 = api_v3
/ = admin_version_api

Keystone.conf ↴

[DEFAULT]
admin_token = c01f11a39faceecde032
[assignment]
[auth]
[cache]
[catalog]
driver = keystone.catalog.backends.sql.Catalog
[cors]
allowed_origin = http://172.18.1.199:3000
allow_methods = GET,POST,PUT,DELETE,OPTIONS
allow_headers = Content-Type,Cache-Control,Content-Language,Expires,Last-Modified,Pragma,X-Auth-Token,X-Subject-Token
expose_headers = Content-Type,Cache-Control,Content-Language,Expires,Last-Modified,Pragma,X-Auth-Token,X-Subject-Token
[cors.subdomain]
[credential]
[database]
connection = mysql+pymysql://keystone:***********************@127.0.0.1/keystone
[domain_config]
[endpoint_filter]
[endpoint_policy]
[eventlet_server]
bind_host = 172.19.1.11
public_bind_host = 172.19.1.11
admin_bind_host = 172.19.1.11
[federation]
[fernet_tokens]
[healthcheck]
[identity]
driver = keystone.identity.backends.sql.Identity
[identity_mapping]
[kvs]
[ldap]
[matchmaker_redis]
[memcache]
[oauth1]
[oslo_messaging_amqp]
[oslo_messaging_kafka]
[oslo_messaging_notifications]
[oslo_messaging_rabbit]
[oslo_messaging_zmq]
[oslo_middleware]
[oslo_policy]
[paste_deploy]
[policy]
[profiler]
[resource]
[revoke]
[role]
[saml]
[security_compliance]
[shadow_users]
[signing]
[token]
provider = fernet
[tokenless_auth]
[trust]

We are using Ocata versoin.

Would you please help??

Hi there from Grafana Labs 👋

Firstly, thank you for contributing to the Grafana plugin ecosystem.

I'm reaching out as this plugin is at risk of being deprecated from the Grafana plugin catalog as it depends on Angular (#83) and has not been updated in a number of years.

We recognise that priorities change and that it may no longer be possible to keep plugins up to date. For your awareness, deprecation has no impact on existing users of the plugin, but discourages further adoption to prevent users from starting to rely on a component which won't receive updates.

Please advise if you plan to migrate, or alternatively that you agree the plugin should be deprecated. In the case of the latter, if there is any specific guidance you would like to offer users of the plugin - please let me know and we can include this in the deprecation notice.

Feel free to email me at [email protected] if preferred; apologies, we don't have an email address associated with this plugin so I had to raise an issue.

If we don't hear from you within 30 days, I'll go ahead and deprecate the plugin.

Thank you again
David

Hello 👋 from Grafana Labs

We have detected that your plugin might still be using AngularJS and recommend migrating it to React, so that it may continue to operate successfully in future versions of Grafana.

You can read more on why and how by visiting our documentation: https://grafana.com/docs/grafana/latest/developers/angular_deprecation/

You can reach out to us with any questions via slack: slack.grafana.com - channel #plugins or our plugin development community forum

Hi,

Didn't change the settings, but after upgrade to grafana 7, the datasource just seems not to be working. It was working previously with keystone credentials set.

When I hit "Save and test" the blue bar shows "Testing" forever.

in the grafana logs, I just see a 200..

t=2020-06-09T16:52:40+0200 lvl=info msg="Request Completed" logger=context userId=67 orgId=1 uname=<email> method=GET path=/api/frontend/settings status=200 remote_addr=123.123.123.123 time_ms=5 size=21483 referer=https://grafana-url.url/datasources/edit/19/

any advise or hints to track it down?

For example I have an aggregation of all "cpu-X" metrics of one resource.

The metrics() and resources() variables do not appear in the Templating Variables list. They do not function in the queries.

Your plugin documentation:

Templated queries
Gnocchi Datasource Plugin provides following functions in Variables values query field in Templating Editor.

Name Description
metrics(resource_id) Returns a list of metrics avialable for the resource identified by ‘resource_id’
resources(resource_type,attribute, query) Returns a list of resource attribute matching query.
For details of query format, please refer to the Gnocchi documentations.

If I have the wrong expectations for these to appear automatically, can you provide an example of how to properly use these? I'd like a template variable for all resources that are active in gnocchi within the specified time frame.

Problem Statement

Getting an error when selecting Save & Test Gnocchi data source stating: Keystone failure: (404 Not Found).

The associated error was found on the openstack controller in /var/log/keystone/keystone_access.log:

<ip redacted> - - [27/Nov/2018:14:21:58 +0000] "POST /v3/v3/auth/tokens HTTP/1.1" 404 350 "-" "Grafana/5.3.4"

An example of a correctly formed URI looks like:

<ip redacted> - - [27/Nov/2018:14:49:58 +0000] "POST /v3/auth/tokens HTTP/1.1" 201 10038 "-" "python-openstackclient keystoneauth1/2.4.1 python-requests/2.9.1 CPython/2.7.12"

Note: An extra keystone version declaration is found in the access log associated to the error.

Keystone endpoint on controller

The endpoint on the controller already contains the version in the URL

  {
    "Service Type": "identity",
    "URL": "http://pod3controller-int:5000/v3/",
    "Region": "RegionOne",
    "Enabled": true,
    "Interface": "internal",
    "Service Name": "keystone",
    "ID": "9c5ad5235f7d4fc1a1b2d6b0a2b72230"
  }

TCP Conversation.

This is the TCP conversation as seen from the controller filtered on only the grafana host

POST /v3/auth/tokens HTTP/1.1
Host: pod3controller:8041
User-Agent: Grafana/5.3.4
Content-Length: 196
Accept: application/json, text/plain, */*
Accept-Encoding: gzip, deflate
Accept-Language: en-US,en;q=0.5
Content-Type: application/json
X-Forwarded-For: <IP redacted>, <IP redacted>
X-Grafana-Nocache: true
X-Grafana-Org-Id: 1

{"auth":{"identity":{"methods":["password"],"password":{"user":{"name":"admin","password":"<redacted>","domain":{"id":"Default"}}}},"scope":{"project":{"domain":{"id":"Default"},"name":"admin"}}}}HTTP/1.0 404 Not Found
Date: Tue, 27 Nov 2018 17:01:18 GMT
Server: WSGIServer/0.1 Python/2.7.12
Content-Length: 112
Content-Type: application/json; charset=UTF-8

{"message": "The resource could not be found.<br /><br />\n\n\n", "code": "404 Not Found", "title": "Not Found"}

Environment

Installed Grafana 5.3.4 (commit: 69630b9) by following the installation document for apt repository on a virtual instance of Ubuntu 16.04.5 LTS (xenial)

Installed the plugin following the Gnocchi plugin for Grafana

Datasource Config

Request

Anyone know why this extra version declaration is present in the plugin's keystone request for a token and how to correct it?

I'm not able to use the Keystone Auth Mode.

Even if the services are using the same URL, they don't share the same port.
As a result, I can't use the keystone configuration, but only the token one.
It may be the same problem encountered in #68.

I create a PR #70 to implement it.

Hello. I'm trying to configure the Gnocchi Grafana plugin to access a lab install of Pike deployed using OpenStack-Ansible. By default, the deploy uses self-signed certs and the public Keystone VIP is automatically https. Since there is no Horizon plugin for metrics (that I'm aware of), it looks like this the winner for visualizing the metrics. The deploy seems to be working properly in all respects that I've tested including executing openstack CLI commands from an external server using the public endpoints. (For example, "openstack server list --insecure").

I can't seem to get anything to work with the Grafana plugin on an external server. It could be I'm just doing something simple incorrectly. I'm primarily testing the proxy method and receiving "Keystone failure: (502 Bad Gateway)". I'm testing against an OpenStack Pike environment using OpenStack Ansible 16.0.7.

Pretty much any combination of things I try result in some form of 502 keystone bad gateway error.

An example of settings I'm trying currently are seen in the attached image:

Endpoints as seen from within the OS environment in the attached image:

If you can offer any pointers, please advise.

Thank you!

When crafting queries in the Grafana UI, it would be helpful if the names of available metrics were discoverable from within the "metric regex" field, rather than requiring an out-of-band connection in order to discover available metric names.

DataSource add normally,Specify that data will not be displayed after data source is imported into dashboard(id 4569).

When selecting many (or all) instances, it will try to do a request to gnocchi, including all the instance ids as filters in the query. This will fail with Failed to load resource: the server responded with a status of 414 (Request-URI Too Long).

This should probably be decomposed in multiple requests.

I found that enabling cors for gnocchi in the latest RDO requires the gnocchi api-paste.ini pipeline to look like:

[pipeline:main]
pipeline = cors gnocchi+auth

Hello,

I'm working on a setup on centos 7 with the following packages (gnocchi packages from rdo-ocata and grafana from grafana.net):

grafana-4.4.3-1.x86_64
openstack-gnocchi-api-3.1.9-1.el7.noarch
openstack-gnocchi-common-3.1.9-1.el7.noarch
openstack-gnocchi-indexer-sqlalchemy-3.1.9-1.el7.noarch
openstack-gnocchi-metricd-3.1.9-1.el7.noarch
python-gnocchi-3.1.9-1.el7.noarch

And I installed collectd-gnocchi (1.5.0) to get system metrics.

I installed 'gnocchixyz-gnocchi-datasource @ 1.2.0' with grafana-cli tool

I created a datasource (without keystone auth) and got "Success Data source is working"

When I tried to create a graph, I got the following error:

Missing or invalid fields: Query, Metric name when creating a dashboard

Do you have any idea how to fix this issue ?

Thanks

Including the ad-hoc templates

Hi! Please tell me, I use the latest version of grafana 9.4.3, unfortunately
I can't connect to the data source I use keystone auth mode and the test lasts indefinitely,
if I use noauth mode I get starus code 400, can you help me figure out what's the matter,
why it doesn't work?

Graphs should include units for metrics. It is hard enough to identify what a metric represents other than manually looking up the UUID with the gnocchi cli. The units would help us understand what the values illustrate.

It seems as if the "resource type" field (when using a "resource search" query) ought to be a drop-down menu. It seems that in general there will be a relatively small number of resources types, and you can always use a combo box if you still want people to be able to type in an arbitrary value.

the credential is passed to keystone direct from browser and even non-grafana-admin user can get the password

I would like to use the metric name as the label in grafana graphs. When I enter name into the "Label" field, I see in the graph "attribute name not found". On the other hand, other attributes (like "creator") work fine. I'm confused, because gnocchi metrics clearly have a name attribute. Am I doing something wrong?

Having a dashboard with range Last 24 hours.

Doing a resource search with query display_name = '[[server]]', returns me 9 resources (instances).
All of them are loaded, but only one has data.
On each resource, ended_at should be check if is null (not ended) or in range of time -
started_at, in order to avoid loading the other resources with no data at that time.

I've been using grafana-gnocchi-datasource token based authentication with a small hacky service that runs regularly and updates the token in the database. https://gitlab.com/Baggypants/grafana-gnocchi-openstack-token-renewer.

It would be helpful if there was a mechanism in the datasource to reload the token from the grafana database at a given period. For the time being I'm using a tab auto refresh browser plugin.

Hello!

I installed gnocchi and ceilometer using devstack mitaka . When I go to grafana dashboard and add datasource from gnocchi using direct mode.

It has an error:

Gnocchi error: No response status code, is CORS correctly configured ?

Please check for me.

Resources and metrics identified by UUIDs are meaningless. The plugin should be able to show the human-readable name. This will simplify query creation and make the graphs more meaningful.

See https://bugs.launchpad.net/gnocchi/+bug/1684047

While attempting to add this plugin I get the error:
Gnocchi error: No response status code, is CORS correctly configured ? (detail: [object Object])

I find that this originates in dist/datasource.js:815 in _gnocchi_auth_request which reads:
reason.message = "Gnocchi error: No response status code, is CORS correctly configured ? (detail: " + reason + ")"

Can this be changed to reason.status or something more useful for trouble shooting CORS problems?

Hello,
we are using the Gnocchi Grafana datasource v1.6.2 and are trying to query the metrics of an OpenStack cluster. We noticed that the dynamic aggregate for the instance metric memory returns weird values. If you compare the resource search method using metric regex memory with dynamic aggregates using the Operation(metric memory mean), the result will be very different. Only the resource search method returns the correct memory allocation for each instance while the dynamic aggregate just returns zeros