[FIXED] 1.1.1.1/help is broken and often shows wrong diagnosis

Almost every time I load https://1.1.1.1/help I get a different result. Most of the times it does not show me connected to 1.1.1.1, even though I most certainly am.
I am using 1.1.1.1 solely via DNS over HTTPS with both IPv4 and IPv6 connectivity.

Result 1 results in “no connectivity to 1.1.1.1 and no DoH” even though the AS number clearly shows Cloudflare.
Result 2 shows that I am connected to 1.1.1.1 via DNS over HTTPS, however I am not actually connected to 1.1.1.1 :doge: huh? AS number is also still Cloudflare.
Result 3 shows the correct and expected result with a connection to 1.1.1.1, using DoH.

Could this be looked into?

Each of those is a separate lookup + fetch, so it can be inconsistent if you have a DoH fallback configured in the browser/service you’re using, or if you have something in between the browser and 1.1.1.1. You should be able to see from the web developer tools which assets are not loading.

I have no fallback resolver nor is there any relay happening. It also happens on WARP on my phone or when using normal DNS. “Connected to 1.1.1.1” often just says No.

I noticed that the request to https://91864fe0-d133-42e4-8938-1385c92417eb.is-cf.help.every1dns.net/resolvertest times out, so the result will be No.
Opening the page in the browser shows this on first try:


Refreshing the page shows “1”, which should be the first expected response.

Thanks that’s super helpful, the name does resolve but seems like something is going on with the website. I’m not able to reproduce it so it might be local. Let me add a ticket to track this.

Have been experiencing the same issue for a while now. If I do a test I have to run about 5 or 6 times to get a decent result. Every time seems that some of the checks just fail. I.e. can be connected via WARP o the desktop, can see the DNS queries passing over DoH or WARP in the log, however website will say No to everything (including connected to 1.1.1.1) - Few refreshes later and it seems to be correct.

Hi @milk and @hdsnf723, we’ve made some changes recently, can you check if the issue still exist on your side? Thanks.

@anb
Wow, the check is now very fast! Before it used to hang for 5-10 seconds on “Checking…”, now it’s near instant.
“Connected to 1.1.1.1” now always seems to be correct, which is great. I did notice, however, that when using WARP, it shows that I’m using DoH and not WARP. Furthermore, when on an IPv4 only connection, it shows that there is a connection to the IPv6 addresses – which is impossible.

I noticed that the IPv6 test is being cached from disk sometimes:


This half IPv6 result happened to me after being on an IPv6 network, running the test, connecting to an IPv4-only network and re-running the test. There also seems to be a Cache-Control header present on the resolvertest resource, perhaps changing this to Cache-Control: no-store, max-age=0 will fix the issue. @anb

Oops, my bad. I was experimenting with different settings and switched on DNS over HTTPS instead of DNS over WARP in the 1.1.1.1 app. That’s why DoH was showing on the debug page, and not DoWARP when using WARP. Understandable, but I would find it more logical to have DoHTTPS and DoWARP both marked as Yes when using DoH over WARP. So the debug page has at least one indicator that WARP is working correctly, even when using a different DNS transport.

I see the Cache-Control: no-store header has been added now. The test now consistently works correctly even on a soft refresh, great! I suppose this is now resolved.

1 Like

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.