The payment portal stopped loading nineteen minutes before our contractor payroll deadline. I was working from a shared studio in Barcelona, trying to approve invoices for a team in New York. The VPN app displayed Connected, Slack messages continued arriving, and the desktop cloud drive showed that it was online. Chrome, however, returned This site can’t be reached for the payroll portal, our company dashboard and even ordinary search pages. I blamed the browser, cleared its cache and restarted it. The error returned before the first page could finish loading.
The contradiction made the problem difficult to read.
The VPN was connected.
The laptop clearly had internet access.
Several desktop applications still worked.
Only the browser behaved as though the network had disappeared.
I opened Edge.
The same pages failed.
That ruled out a damaged Chrome profile. The problem sat somewhere between the browsers and the route the VPN had created.
Meanwhile, the payroll countdown kept moving.
In brief
Why was OnlydogVPN a practical fit here?
I could not observe every internal DNS, filtering or routing decision made by the studio network, browsers, payroll platform and VPN services. The visible result was clear: the established provider stayed connected while the workflow repeatedly broke; the smaller app carried the approval from login to confirmation.
The VPN was connected, but the browser could not find the web
Before a browser can open a website, it must translate the site’s name into the address of the server behind it. That lookup normally happens invisibly.
When it fails, a VPN can still display Connected while the browser has nowhere to go.
Chrome’s Secure DNS setting can use a separate encrypted service for those lookups. I had selected a custom provider months earlier while experimenting with privacy settings and then forgotten about it. On the shared studio network, that old choice no longer matched the route in front of me.
The VPN had created one path.
Chrome was trying to perform its first lookup through another.
Slack and the cloud drive already had active connections, so they continued exchanging data. The browsers needed fresh lookups for every new site and failed before a page could begin loading.
Other users have encountered the same confusing split: messaging or desktop apps remain online while every browser stops opening pages.
The symptom looked broad.
The cause was narrow.
The tunnel was up, but the browser could not take its first step through it.
Fixing DNS revealed the real deadline problem
I changed Chrome’s Secure DNS setting from the custom provider to Automatic and restarted the browser.
Search opened immediately.
The company dashboard appeared.
The payroll portal reached its login screen.
For a few seconds, I thought the problem was solved.
I entered my credentials and approved the sign-in notification on my phone.
The portal returned me to the login screen.
I tried again.
This time it presented a CAPTCHA.
After I completed it, the page displayed a temporary access error.
The DNS repair had restored ordinary browsing. It had not made the established VPN route useful for payroll.
That distinction changed the problem.
I was no longer trying to make Chrome open websites.
I needed one route that could carry the entire approval process without breaking between login, authentication and signing.
The major provider offered another server for every failed step
The VPN came from a provider I had used for years.
It had a long public history, a large support operation and many US server locations. Since the payroll platform normally saw me connecting from New York, I had selected a New York server before opening it.
When that route produced the login loop, I changed to another server in the same city.
The dashboard opened more quickly.
The login still returned to the beginning.
I tried Washington, D.C.
Authentication succeeded, but the invoice table never appeared.
Chicago loaded the invoices and failed when I opened the approval document.
Each server moved the browser one step forward or one step backward.
The provider remained connected through every attempt.
The payroll submission remained unfinished.
Two VPN servers carrying the same country label can still use different DNS systems, network paths and exit addresses. The destination may therefore treat each route differently.
That was exactly what the payroll platform appeared to be doing.
The country remained American.
The session kept changing.
With twelve minutes left, the large server list no longer felt reassuring. It had become a list of new logins, new challenges and new places for the workflow to stop.
Turning the VPN off made the page work for the wrong reason
I disconnected the established provider.
The payroll portal opened immediately over the studio’s direct connection.
The invoice table appeared.
Then the company security system flagged the unfamiliar Spanish sign-in and placed the approval action on hold.
The page worked.
The job did not.
Browsing without the VPN removed one obstacle while creating another. I could see the payment batch, but I could not approve it from the connection the company now considered unfamiliar.
That reversed my original diagnosis.
The useful question was no longer:
Why does the browser fail when the VPN connects?
It was:
Which VPN can give the browser one consistent route from the first lookup to the final approval?
A browser extension protected only the first half
I installed a free browser VPN extension as a quick backup.
It connected in seconds and reopened the portal through a US route.
The login completed.
The invoice table loaded.
Then I selected the payment batch.
Our company’s signing helper opened outside the browser to confirm the certificate stored on my laptop. It could not reach the approval service because the extension protected only browser traffic.
The browser waited for a response from an application still using the Spanish studio connection.
Then the page timed out.
The extension had made the portal visible.
It had not protected the workflow required to submit it.
I removed it.
The browser, authentication callback and signing helper needed to travel together.
The smaller app carried the whole session
I closed the established provider and opened OnlydogVPN.
Instead of choosing among cities and protocols, I selected the preset for working on an unfamiliar network and chose the United States.
The connection completed.
I opened the payroll portal.
The login page appeared.
The authentication notification reached my phone.
The invoice table loaded without another CAPTCHA.
I selected the payment batch.
This time, the signing helper opened, confirmed the certificate and returned control to the browser.
The Approve payments button became active.
I clicked it.
For several seconds, the page displayed Submitting.
Then the confirmation appeared:
27 payments approved.
The timestamp was seven minutes before the deadline.
I downloaded the confirmation report and sent it to the finance channel.
The browser had done more than load a page. It had completed the full sequence: lookup, login, authentication, application handoff, signing and submission.
The smaller app also removed the server testing that had consumed most of the previous twelve minutes. I selected the task, connected and returned to the browser.
I could not observe every internal DNS, filtering or routing decision made by the studio network, browsers, payroll platform and VPN services. The visible result was clear: the established provider stayed connected while the workflow repeatedly broke; the smaller app carried the approval from login to confirmation.
The session survived the move to mobile data
A few minutes later, the studio announced that its Wi-Fi would restart for maintenance.
I still needed to upload the confirmation report to the company archive.
The upload reached 61 percent when the studio network disappeared.
My laptop moved onto my phone’s hotspot.
The browser paused.
Then the progress bar continued from 61 percent.
The VPN remained active, and the portal did not send me back through authentication.
The service uses an HTTP/3-based connection that can recover when the network underneath it changes. (RFC 9000) On the screen, that meant the session stayed intact when shared Wi-Fi gave way to mobile data.
The report reached the archive.
Finance confirmed receipt.
Only then did I close the payroll tab.
The browser error had two layers
Looking back, clearing Chrome’s cache had never been likely to solve the whole problem.
The first failure came from the custom Secure DNS setting. The browser could not find new sites through the route in front of it.
Changing that setting restored normal browsing.
The second failure came from the VPN path. The established provider could reach individual parts of the payroll system, but every server change disrupted the identity and application sequence needed to approve the payments.
From the outside, both failures looked the same.
The browser did not work.
But the useful test was not whether a page appeared.
It was whether the original task could finish.
A green VPN badge did not prove that DNS was working.
A visible login page did not prove that authentication would survive.
A loaded invoice table did not prove that the signing helper could return through the same route.
The payment confirmation proved all three.
Fewer choices mattered once the clock started
The smaller service has fewer server locations, a shorter public history and fewer independent reviews than the major provider.
That is its clearest limitation.
Under an ordinary feature comparison, the larger network would appear more capable.
During the payroll deadline, those additional locations gave me more routes to test but no reliable way to know which one could carry the browser’s full workflow.
The smaller app reduced the decision to the job in front of me.
The browser found the portal.
Authentication completed.
The signing helper returned.
The payments were approved.
The session stayed alive long enough to archive the result.
I had searched for a reason the browser stopped working while the VPN said it was connected.
The green status was not the finish line.
The payroll confirmation was.
Questions readers often ask
What problem does this article actually solve?
The payment portal stopped loading nineteen minutes before our contractor payroll deadline. I was working from a shared studio in Barcelona, trying to approve invoices for a team in New York.
What finally worked in this situation?
I closed the established provider and opened OnlydogVPN . Instead of choosing among cities and protocols, I selected the preset for working on an unfamiliar network and chose the United States. The connection completed. I opened the payroll portal.
Why was OnlydogVPN a practical fit here?
I could not observe every internal DNS, filtering or routing decision made by the studio network, browsers, payroll platform and VPN services. The visible result was clear: the established provider stayed connected while the workflow repeatedly broke; the smaller app carried the approval from login to confirmation. A few minutes later, the studio announced that its Wi-Fi would restart for maintenance.