You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I believe there may be a regression related to the country detection issue previously discussed in #3583 and reportedly fixed in v3.
Today I am observing a significant number of visitors being reported as:
Country: Poland
Region: Mazowieckie
City: Warsaw
However, multiple independent signals suggest this geolocation is incorrect.
The pattern appears repeatedly across different browser fingerprints and devices and is disproportionately concentrated in Warsaw.
Environment
Umami Cloud
Current Cloud version
No self-hosted instance
Standard Umami Cloud integration
Script served from first-party endpoint (/script.js)
Analytics payloads sent directly to api-gateway.umami.dev
Why I Suspect a Problem
Issue #3583 described a situation where country detection on Umami Cloud differed from self-hosted deployments due to IP/header handling and proxy-related behaviour.
The issue was eventually marked as fixed in v3:
Fixed in v3.
However, the behaviour I am observing today appears similar:
A large cluster of traffic is consistently attributed to Warsaw, Poland.
The attribution appears implausible given the site's audience profile.
The pattern is persistent and repeatable.
Questions
How does Umami Cloud currently determine visitor country?
Does Umami Cloud still prioritise cf-connecting-ip over other headers?
Are there any known issues with geolocation accuracy for Cloud customers in 2026?
Is there a way to inspect the raw IP geolocation result used for a specific session?
Additional Context
The issue is not that the traffic exists.
The concern is that the country attribution itself appears incorrect and resembles the behaviour previously described in #3583
Given that a similar issue existed previously and was reportedly fixed in v3, I would like to confirm whether the Cloud deployment is still using the corrected logic and whether any regressions are known.
Describe the Bug
I believe there may be a regression related to the country detection issue previously discussed in #3583 and reportedly fixed in v3.
Today I am observing a significant number of visitors being reported as:
However, multiple independent signals suggest this geolocation is incorrect.
The pattern appears repeatedly across different browser fingerprints and devices and is disproportionately concentrated in Warsaw.
Environment
/script.js)api-gateway.umami.devWhy I Suspect a Problem
Issue #3583 described a situation where country detection on Umami Cloud differed from self-hosted deployments due to IP/header handling and proxy-related behaviour.
The issue was eventually marked as fixed in v3:
However, the behaviour I am observing today appears similar:
Questions
cf-connecting-ipover other headers?Additional Context
The issue is not that the traffic exists.
The concern is that the country attribution itself appears incorrect and resembles the behaviour previously described in #3583
Given that a similar issue existed previously and was reportedly fixed in v3, I would like to confirm whether the Cloud deployment is still using the corrected logic and whether any regressions are known.
Thank you.
Database
Umami Cloud
Relevant log output
Which Umami version are you using?
No response
How are you deploying your application?
No response
Which browser are you using?
Chrome