Convert Istio dashboards to Kibana Lens #5268

gsantoro · 2023-02-14T12:23:38Z

What does this PR do?

Convert the Istio dashboards to Lens

Checklist

I have reviewed tips for building integrations and this pull request is aligned with them.
I have verified that all data streams collect metrics or logs.
I have added an entry to my package's changelog.yml file.
I have verified that Kibana version constraints are current according to guidelines.

Author's Checklist

[ ]

How to test this PR locally

Related issues

Relates Istio metrics and dashboards #4678
Closes Add data_stream.dataset pre filter in Istio dashboards #4978

Screenshots

elasticmachine · 2023-02-14T12:37:03Z

💚 Build Succeeded

the below badges are clickable and redirect to their specific view in the CI or DOCS

Expand to view the summary

Build stats

Start Time: 2023-03-06T20:56:41.791+0000
Duration: 15 min 44 sec

Test stats 🧪

Test	Results
Failed	0
Passed	14
Skipped	0
Total	14

🤖 GitHub comments

Expand to view the GitHub comments

To re-run your PR in the CI, just comment with:

/test : Re-trigger the build.

constanca-m · 2023-02-14T13:26:57Z

Seems ok. For the jenkins error, update the kibana requirement here to 8.6.0, or change the version here and here for the dashboard (8.4.0 if possible).

gizas · 2023-02-15T15:32:19Z

packages/istio/kibana/dashboard/istio-f0f863b0-5941-11ed-bfb8-bbfe5b338339.json

+                                                    "dataType": "number",
+                                                    "isBucketed": false,
+                                                    "label": " ",
+                                                    "operationType": "max",


Should this be max?

About this metric. I don't even know what it's doing but it is not one of those ones reported by Stephen to convert from Max to Sum. Before I change that, I'll have a chat with Stephen.

gizas · 2023-02-15T15:33:27Z

packages/istio/kibana/dashboard/istio-f0f863b0-5941-11ed-bfb8-bbfe5b338339.json

-                "title": "Galley Validation Failed",
-                "type": "visualization",
-                "version": "8.4.0"
+                "title": "Galley Validation Failed by Label",


Just tempted to ask if we need Galley in general. Do you think it gives value? Do you think we can remove it?

I have no idea what it is yet. @ChrsMark do you know?

Istiod still reports this metric, right? If so, why to remove it?
The definition of this metric seems to be Count of validations failed by the configuration validation webhook. The metric is reported as galley_validation_failed{group="networking.istio.io",reason="invalid_resource",resource="gateways",version="v1alpha3"} 1.

So I would say if it's still provided, it's a valuable indicator that Istio provides and some other tools use it to create alerts. See an example here. So I wouldn't remove it if there is room for it in the dashboard but it's up to you folks to make the last call.

I'll have a look at that metric and I'll write here about it.

gizas · 2023-02-15T15:48:45Z

We can keep adding short info for how we test and also screenshots to provide evidence for your results.
Dashboard stories with screenshot evidence should be a must.

Also in general the description text like Metrics from Istiod does not give much value. Either remove it or enhance it with valuable information on how the users should read the dashboards .

Some more ideas like grouping Pilot XDs together or Time related visualisation can also be handy. Especially in overview I am getting lost to read so many information.

Also @constanca-m used markdown to group specific visualisations. You can also group by this your visualisations like here elastic/beats#34161

Overall great improvement Giuseppe!

gsantoro · 2023-02-15T16:32:03Z

We can keep adding short info for how we test and also screenshots to provide evidence for your results. Dashboard stories with screenshot evidence should be a must.

I agree. I opened this PR in draft mostly to see what was broken with the Jenkins build before adding screenshots and instructions to test.

Also in general the description text like Metrics from Istiod does not give much value. Either remove it or enhance it with valuable information on how the users should read the dashboards .

I am not sure what to put in there. I just wanted to convey that those metrics in Overview are from Istiod compared to the ones in Traffic which have a description Metrics from Istio Sidecar Proxies. I am open to suggestions.

Some more ideas like grouping Pilot XDs together or Time related visualisation can also be handy. Especially in overview I am getting lost to read so many information.

Also @constanca-m used markdown to group specific visualisations. You can also group by this your visualisations like here elastic/beats#34161

Looks good. I'll have a look into those. Thanks

Overall great improvement Giuseppe!

gsantoro · 2023-02-15T17:15:00Z

@constanca-m and @gizas

Also @constanca-m used markdown to group specific visualisations. You can also group by this your visualisations like here elastic/beats#34161

It looks great that piece of markdown. Unfortunately it's a TSVB panel and the styling with custom CSS is not available in Lens. Given that the topic of this PR is to move away from TSVB I am not sure that I should add a TSVB panel for styling some markdown.

elasticmachine · 2023-02-15T17:44:50Z

🌐 Coverage report

Name	Metrics % (`covered/total`)	Diff
Packages	100.0% (`1/1`)	💚
Files	100.0% (`3/3`)	💚
Classes	100.0% (`3/3`)	💚
Methods	90.0% (`27/30`)	👍 65.0
Lines	97.472% (`347/356`)	👎 -2.528
Conditionals	100.0% (`0/0`)	💚

gizas · 2023-02-16T10:03:47Z

Ok markdown in TSVB is not something we must do it. Just group then the related visualisations
For suggestions for text: You can even put links to Istio pages that describe the components you visualise for reference.
Keep Galley
Have a look for the max comment Convert Istio dashboards to Kibana Lens #5268 (comment)

gsantoro · 2023-02-16T10:27:56Z

I would like to start a discussion about the changes applied to migrate from TSVB to Lens. I'll highlight every change in a separate comment so that we easily reference them and comment them.

in Lens you can't see the last value for a metric in the Legend. The current value is displayed as usual when you over the metric.

In TSVB

in Lens

gsantoro · 2023-02-16T13:27:10Z

Requests per source/destination used to use the Max function instead of Sum function. . Now instead, we sum the number of requests per source/destination and we calculate the rate in seconds as well. They have migrated to Lens as well

old implementation in TSVB with Max function

new implementation in Lens with Sum function

gsantoro · 2023-02-16T13:30:29Z

Size of requests/responses has been rewritten from TSVB to Lens. (99) for 99th percentile in the legends has been replaced by (99th percentile) in the titles

in TSVB

in Lens

gsantoro · 2023-02-16T13:31:42Z

Similar discussion as point 3 for Duration of requests by source/destination

in TSVB

in Lens

gsantoro · 2023-02-16T13:36:03Z

Multiple Gauges being replaced by Numeric Metric colour coded in 3 ranges (>0 and < 50%, >= 50% and < 75%, > 75% and <100%) with colours Green, Yellow, Red similarly to the behaviour with Gauges

in TSVB

in Lens

gsantoro · 2023-02-16T13:37:54Z

Multiple percentiles in the same visualization

in TSVB (notice the last value after the metric is a bit confusing here

In Lens (percentiles are indicated with (25) for 25th percentile, ...

gsantoro · 2023-02-16T13:42:08Z

The final look of both dashboards Overview and Traffic

in TSVB

in Lens

gsantoro · 2023-02-16T13:42:56Z

Hey @bvader,
do you mind having a look at this PR for the changes that we discussed on our last meeting?

gsantoro · 2023-02-16T16:02:36Z

I recently saw that Kubernetes integration comes with a new way to implement Gauges (like in the below screenshot).

The problem with that is it's coming from a Gauge TSVB visualization.

Since there was lot of focus on moving the Istio integration entirely to Lens, I am now wondering how much of hour current dashboards still rely on TSVB given that I have seen at least

markdown for category of dashboards
Gauges for percentage usage
(there might be more I am not aware of yet)

gizas · 2023-02-20T09:12:32Z

At the moment all above changes LGTM.
Two minor:

Foot Labels like @timestamp per hour can be removed? Or to be included in the header of visualisation?
Numeric Metrics in 5: Can you change the actual number to be in the center and big? Also Capitialise headers and make them onliners?

gsantoro · 2023-02-21T14:48:10Z

At the moment all above changes LGTM. Two minor:
thanks for the feedback

Foot Labels like @timestamp per hour can be removed? Or to be included in the header of visualisation?

About the foot label. I can either remove the label entirely or customize the @timestamp part. The second part of that label actually change based on the timeframe. So that part might be per hour or per seconds. I can customize that part other than fixing the time frame or remove that part entirely.

this is how you remove the @timestamp part of that label

and this is how it looks with no X axis label.

What do you suggest?

Numeric Metrics in 5: Can you change the actual number to be in the center and big? Also Capitialise headers and make them onliners?

I can't modify any of the visual aspect of this Metric visualization, meaning either the aspect of the number or the headers. TSVB has lot more customization capabilities (via CSS) than Lens. The only alternative is to have a single big percentage number but we will be getting rid of the color that indicate a possible range of values.

constanca-m · 2023-02-21T14:55:09Z

It looks great that piece of markdown. Unfortunately it's a TSVB panel and the styling with custom CSS is not available in Lens. Given that the topic of this PR is to move away from TSVB I am not sure that I should add a TSVB panel for styling some markdown.

This is unfortunately true. There are other things that Lens can't do yet, so I wouldn't call having this text as a problem. Example of another thing: summing the last value of some metric that breaks down in a label (Limitations section in this PR)

constanca-m · 2023-02-21T14:59:25Z

Foot Labels like @timestamp per hour can be removed? Or to be included in the header of visualisation?

If it were me, I would try to remove the axis labels, and try to be clear with the title and the legend (if it exists). And there is no reason to add the unit type (example, sec, %) in the title, if we can write the axis value with a suffix - like 4 sec. @gsantoro

gsantoro · 2023-02-21T16:31:06Z

If it were me, I would try to remove the axis labels, and try to be clear with the title and the legend (if it exists). And there is no reason to add the unit type (example, sec, %) in the title, if we can write the axis value with a suffix - like 4 sec. @gsantoro

I am more than glad to remove the X axis entirely, fix the date histrogram resolution (like @gizas was suggesting) and let the user guess the timeframe by moving the cursor through the values. Unfortunately you can't add a suffix to an X axis when the axis has a Date histogram. I have done than when the Y axis was numeric to add a s for seconds.

This is how it looks with no X axis and 1h fixed timeframe

constanca-m · 2023-02-21T16:41:04Z

let the user guess the timeframe by moving the cursor through the values ... This is how it looks with no X axis and 1h fixed timeframe

The visualization is very understandable, I don't see a problem in not having a suffix in the x values.

gsantoro · 2023-02-21T17:33:03Z

Here it is the updated look of the dashboard. Changes are:

inline all pilot Numeric metrics
removed X axis from all visualization
set the default Date Histogram resolution to 1h. Just to be clear. If you have more than a couple of days of time frame, the dashboard use a 3h bucket size instead of 1h. This bucket size changes automatically by default based on the timeframe. Fixing it to 1h might have performance implication if you look at more 1w time frame. I hope everyone is ok with this tradeoff.

constanca-m · 2023-02-21T18:17:00Z

The link to the screenshot just leads me to the description of the PR @gsantoro

gizas · 2023-02-22T08:28:13Z

set the default Date Histogram resolution to 1h. Just to be clear. If you have more than a couple of days of time frame, the dashboard use a 3h bucket size instead of 1h. This bucket size changes automatically by default based on the timeframe. Fixing it to 1h might have performance implication if you look at more 1w time frame. I hope everyone is ok with this tradeoff.

Such kind f information can go to a small markdown if you think the user should know. Sth like Default size of bucket is 1h, you can anticipate performance degradation if large number is set

Overall LGTM

gsantoro · 2023-02-22T09:47:08Z

@constanca-m sorry about that. I fixed the screenshot now.

@gizas I am personally not a huge fan of adding a markdown text to explain the decision behind a choice that might affect the performance of a dashboard. I would have preferred to keep the time frame dynamic like we do elsewhere and then do not required the markdown. Nevertheless, I am happy to go with this solution if you think it is better.

gizas · 2023-02-22T10:05:57Z

Sure no strong opinion on this, just an alternative if you wanted to consider.

gsantoro · 2023-02-22T11:23:52Z

I got some feedback from Stephen Brown that the Istiod Dashboards with 3 visualization per row are a a bit cramped even on a fairly wide screen.

This is how it looks with 2 visualization per row instead.

I agree with him that it looks better now. So I'll keep it like this unless someone object here.

About the time frame being fixed or not from a previous discussion. This is really what I would like to do:

keep the time frame dynamic like per default settings so that it doesn't create buckets that are too small or too big, it doesn't impact the performance of the dashboard and we don't have to add the extra bit of markdown
I will still remove the X axis label @timestamp per 3h
I'll leave to the user to notice the time frame resolution by just moving the mouse over a couple of sample point from each visualization.

Since we don't have strong guidelines on the previous topics, this is what I think it's sensible but feel free to point out to a better way to do this or some documentation.

gsantoro · 2023-03-06T18:07:08Z

A couple of small changes:

@bvader made me notice that we have both "Requests per source/destination" and "Requests / sec per source/destination". So I remove the first 2 in favour of just leaving the last 2.
I have removed the X labels from all dashboards (this was already agreed but I had some still there)
I have added the dashboard pre-filters according to Add data_stream.dataset pre filter in Istio dashboards #4978
I have tested everything with some real Istio data

ChrsMark

LGTM!

elasticmachine · 2023-03-07T10:10:57Z

Package istio - 0.2.4 containing this change is available at https://epr.elastic.co/search?package=istio

* convert dashboards from TSVB to Lens * add data_set.datastream pre-filter * some other layout optimizations

gsantoro added enhancement New feature or request draft Draft labels Feb 14, 2023

gsantoro self-assigned this Feb 14, 2023

gsantoro requested a review from a team as a code owner February 14, 2023 12:23

gsantoro requested review from gizas and devamanv February 14, 2023 12:23

gizas reviewed Feb 15, 2023

View reviewed changes

gizas approved these changes Feb 16, 2023

View reviewed changes

gsantoro added 3 commits February 20, 2023 14:00

modified dashboards

b9c8c67

minor changes

f253a23

change minimum version of kibana to 8.6.0

17702b0

gsantoro force-pushed the feature/istio_lens branch from 5642aea to 17702b0 Compare February 20, 2023 14:02

update version to latest changelog

5f819ea

constanca-m approved these changes Feb 22, 2023

View reviewed changes

gsantoro added 3 commits February 28, 2023 18:28

latest update

aa8844e

updated changelog

595b283

restored original dashboard ids

58ae657

updated dashboards

e7a56c9

gsantoro requested review from gizas, constanca-m and ChrsMark March 6, 2023 20:57

gsantoro mentioned this pull request Mar 6, 2023

Istio metrics and dashboards #4678

Closed

14 tasks

ChrsMark approved these changes Mar 7, 2023

View reviewed changes

gizas approved these changes Mar 7, 2023

View reviewed changes

gsantoro merged commit 3c02fde into elastic:main Mar 7, 2023

gsantoro deleted the feature/istio_lens branch March 7, 2023 09:51

agithomas pushed a commit to agithomas/integrations that referenced this pull request Mar 20, 2023

Convert Istio dashboards to Kibana Lens (elastic#5268)

7a64af1

* convert dashboards from TSVB to Lens * add data_set.datastream pre-filter * some other layout optimizations

agithomas pushed a commit to agithomas/integrations that referenced this pull request Mar 21, 2023

Convert Istio dashboards to Kibana Lens (elastic#5268)

ab0db72

* convert dashboards from TSVB to Lens * add data_set.datastream pre-filter * some other layout optimizations

Convert Istio dashboards to Kibana Lens #5268

Convert Istio dashboards to Kibana Lens #5268

Conversation

gsantoro commented Feb 14, 2023 • edited Loading

What does this PR do?

Checklist

Author's Checklist

How to test this PR locally

Related issues

Screenshots

elasticmachine commented Feb 14, 2023 • edited Loading

💚 Build Succeeded

Build stats

Test stats 🧪

🤖 GitHub comments

constanca-m commented Feb 14, 2023

gizas Feb 15, 2023

Choose a reason for hiding this comment

gsantoro Feb 16, 2023

Choose a reason for hiding this comment

gizas Feb 15, 2023 • edited Loading

Choose a reason for hiding this comment

gsantoro Feb 16, 2023

Choose a reason for hiding this comment

ChrsMark Feb 20, 2023

Choose a reason for hiding this comment

gsantoro Feb 21, 2023

Choose a reason for hiding this comment

gizas commented Feb 15, 2023

gsantoro commented Feb 15, 2023

gsantoro commented Feb 15, 2023

elasticmachine commented Feb 15, 2023 • edited Loading

🌐 Coverage report

gizas commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gsantoro commented Feb 16, 2023

gizas commented Feb 20, 2023

gsantoro commented Feb 21, 2023

constanca-m commented Feb 21, 2023 • edited Loading

constanca-m commented Feb 21, 2023

gsantoro commented Feb 21, 2023

constanca-m commented Feb 21, 2023

gsantoro commented Feb 21, 2023 • edited Loading

constanca-m commented Feb 21, 2023

gizas commented Feb 22, 2023

gsantoro commented Feb 22, 2023

gizas commented Feb 22, 2023

gsantoro commented Feb 22, 2023

gsantoro commented Mar 6, 2023

ChrsMark left a comment

Choose a reason for hiding this comment

elasticmachine commented Mar 7, 2023

gsantoro commented Feb 14, 2023 •

edited

Loading

elasticmachine commented Feb 14, 2023 •

edited

Loading

gizas Feb 15, 2023 •

edited

Loading

elasticmachine commented Feb 15, 2023 •

edited

Loading

constanca-m commented Feb 21, 2023 •

edited

Loading

gsantoro commented Feb 21, 2023 •

edited

Loading