Analytics for plugins.jenkins.io - Matomo?

Contributing Infrastructure
22

I could sponsor this for say a year or so to see if it does what we need. I have registered there but can’t add plugins.jenkins.io because it is already taken :slight_smile:

23

We might also want to setup a proxy.

24

I’m leaning towards self hosting matomo for a few months because I have the capacity and does’t cost me anything and i know how to host it.

So far my only goals are to see if:
a) People are clicking footer links (me)
b) People are using the issues tab (@zbynek and I)
Is there any other questions people want answerered for plugins.jenkins.io?

I’m okay with doing a different longer term though

1 Like
25

Indeed it’s registered on my account :slight_smile:

Do you realise that it’s more than 5milions views per month ?
→ Search performance for plugins.jenkins.io

I am interested for a long term solution because I like from time to time look at those data to understand how the traffic is growing and where the people are coming as it helps me to understand where to deploy servers

26

For me you experiment is fine as long as other people have access to the data.
I would prefer to deploy that service on the Jenkins infrastructure with a clear statement that it’s a best effort. As I said there is a value to collect data on long term to analyze trends.

27

ok, 6 months :slight_smile:

for that amount of money it could have been decently self-hosted, maybe.

28

Indeed data are stored in a database so we could rely on a rds or azure postgresql instance.
It shouldn’t be too hard to create a helm chart for plausible, I am not familiar enough with matomo tell

29

I would say its the other way around. Matomo is just php + mysql. There’s a couple helm charts on artifacthub already. I have a diskless version we use at work I’m going to see if i could share if necessary (it just bakes in config.ini.php).

Edit: they even have a helm chart - https://github.com/matomo-org/docker/tree/master/.examples/helm

Plausible uses this clickhouse db which i’ve never heard of, so i suspect can’t use the hosted rds.

I think thats the number of times its shown up in google search results, not how many times the site has been hit.

30

So I finally got around to playing with this again (thanks to [WEBSITE-786] jenkins-matomo.do.g4v.dev certificate expired 2021-10-03 - Jenkins Jira to reminding me about it).

I’ve update the plugin site to track both hashes (bit after # in the url, so tabs) and explicit javascript events for clicking on tabs.

image
image1892×1385 82.6 KB

Because of caching, it take time for new javascript code to go live on everyones browser, but we are already getting some data.

And then general data for the last few months - Jenkins Plugin Site - Web Analytics Reports - Matomo

I do want to setup a helm chart and get it onto normal jenkins infra, but since the server is free for me, and its barely any resourses, and the data is hecka anonymized i’m not in a rush. I’m gonna let it collect some data while I continue to do OT for work, so hopefully in a few weeks I can revisit this.

31

Right now jenkins.io is using google analytics. Only some of us have access, and those arn’t really the ones wanting to use the data. Alyssa asked me for stats for a single blog page today. I dug up some past emails about not wanting to have data shared that isn’t public.

As I’m starting to move away from board related items and back to infra related items i’m starting to think more about old projects.

I’ve been running https://jenkins-matomo.do.g4v.dev/ on a digitalocean droplet for like 2 years now without any issues at all. It seems pretty light weight. Its privacy focused, and can support public anonymous viewing of stats without leaking private data.

Now that we have stable k8s setup, use a hosted db, and hook up jenkins.io, stories.jenkins.io, wiki, accounts, etc to it to get some usage information?

1 Like
32

Definitely nice to have, maybe integrate it into *jenkins.io infrastructure, if we want to add more sites and dashboards?

33

So things important to me, in order are:

  1. Tracking how many hits a page gets.
  • Basic analytics
  1. Everyone has access to the information.
  • One of the reasons I like matomo is you can set it to anonymous reads, so anyone can see the data
  1. Custom event handling (javascript events)
  • Non url based navigation, like switching tabs or carosels or whatever

I looked into plausible, it looks nice, but hosted version would be limited in users, and only users can see data. Hosted matomo is the same, but as mentioned, matomo can be setup for non user to look at basic data.

Self hosted, plausible could be hooked up to ldap so everyone could have access. Matomo can do ldap/oidc/etc.

34

Thanks, It is helpful.