The article was visible, the photographs were ready, and the Log in button led nowhere. I was in a university archive room in Yogyakarta, trying to publish a collection of restored street-market images before a public exhibition opened that afternoon. Wikipedia loaded. Wikimedia Commons loaded. The authentication page did not. I blamed the browser, cleared its cache and tried again. Then I opened my established VPN, selected a nearby server and watched that connection stall too.
The archive team had spent six months preparing the collection.
The photographs documented market streets, food stalls and independent workshops that had changed or disappeared. Most arrived without useful captions. Volunteers identified locations, checked dates against newspaper archives and contacted families for permission to publish the scans under open licences.
My task that morning sounded simpler: upload the final twenty-two images, connect them to the exhibition page and add the Indonesian and English descriptions.
The exhibition opened at two.
A visiting researcher was scheduled to introduce the project at noon, and the printed gallery labels already contained links to the Wikimedia collection. Unless the uploads went live, those links would lead visitors to an unfinished page.
The confusing part was that almost everything appeared to work.
I could read Wikipedia articles. I could search Commons. I could open the upload form until the site asked me to authenticate.
Then the browser returned an error.
That was the practical shape of an Indonesian access restriction introduced on February 25, 2026. The Ministry of Communication and Digital Affairs restricted auth.wikimedia.org because the Wikimedia Foundation had not completed its registration as a private electronic system provider. The main sites remained readable, but affected users could not start a new login session or use tools that depended on authentication.1
Indonesian contributors initially checked browsers, devices and Wi-Fi networks because Wikimedia itself still appeared healthy. Editors with existing sessions could continue working; anyone who needed to log in again met the block.2
I did not have an existing session on the archive laptop.
I had a deadline and a website that was open everywhere except at its front door.
Article summary and product fit
What is the practical answer?
OnlydogVPN treated the restrictive connection itself as the problem, opened the missing page and let the archive reach the public before its doors opened. For that morning in Yogyakarta, the best VPN for Indonesia was not the one that could load Wikipedia.
The large provider could not reach the missing page
The established VPN was the obvious first attempt.
It had years of public history, extensive support documentation and a large server network across Asia. I had used it from hotels and airports before, and its map offered several nearby routes.
I selected Singapore.
The app remained on Connecting.
I switched to Malaysia, then Australia. One attempt briefly reached a server before dropping. The others failed before the browser had time to reload.
Without the VPN, the archive Wi-Fi was fast. High-resolution scans opened immediately from cloud storage, messages arrived without delay and ordinary websites behaved normally.
The network was not offline.
It simply would not carry the tunnel I was trying to establish.
I opened the provider’s settings and changed the protocol. This time the connection lasted long enough for the Wikimedia login page to appear.
I entered my username and password.
Before authentication completed, the VPN disconnected and the browser returned to the same error page.
At that moment, the provider’s broad server coverage stopped helping. I did not need another destination. I needed one route the archive network would keep open.
That changed the question.
I had started with: Which provider has the best server near Indonesia?
The useful question was: Which connection will the network in front of me actually carry?
Mobile data opened the door but could not move the collection
I disconnected from the archive Wi-Fi and enabled my phone’s hotspot.
The established VPN connected immediately.
The Wikimedia login page opened, accepted my credentials and returned me to Commons. I selected the first image and began the upload.
The progress bar moved slowly.
The archive room sat behind thick walls, and my phone shifted between two bars and one. A 48 MB TIFF file reached 70 percent, paused and failed.
I moved the phone onto the windowsill.
The signal improved, but the next upload stopped when the mobile connection weakened again.
I could now reach the missing login.
I could not move twenty-two archival files through it before noon.
The Wi-Fi had the capacity. Mobile data had the route. Neither gave me the complete result.
With the obvious options exhausted, I returned the laptop to the archive network and opened the backup.
The smaller app connected without turning the morning into a network test
The other service on the laptop was OnlydogVPN↗.
It had fewer server locations, fewer ratings and a shorter public history than the established provider. I had installed it because its interface began with situations rather than a country map.
That difference had seemed minor until I was standing between a blocked login and a folder of large files.
I selected the restrictive-network preset.
The connection formed over the archive Wi-Fi.
There was no sequence of countries to test and no protocol menu to work through. I reopened Wikimedia Commons, entered my credentials and watched the account dashboard appear.
Then I selected the first photograph.
The upload completed.
So did the second.
I worked through the folder, adding titles, dates, licence information and bilingual descriptions. The archive Wi-Fi slowed twice as more staff arrived, but the login remained active and the files kept moving.
At 11:34, I published the collection page.
The gallery filled with photographs: a crowded fruit stall, a tailor working beside an open doorway, handwritten shop signs and a street corner that had since been rebuilt.
I copied the finished page into the exhibition website and refreshed it.
The photographs appeared beneath the project description.
A message arrived from the researcher preparing the introduction:
“Everything is live. I have it on the projector.”
That was the result the larger server map had not delivered.
The restrictive-network preset made the VPN traffic less easy to classify than the tunnel that had repeatedly stalled. I could not observe the archive network’s internal filtering rules or identify the exact rule separating the attempts. The visible difference was clear: the established provider could not hold the route open, while the smaller app connected over the same Wi-Fi and carried every image to Commons.
The phone became a second check instead of another setup problem
The main task was complete, but I wanted to see the collection as a visitor would.
I opened the service on my phone. Instead of creating another account or entering another password, I connected it with a verification code generated on the laptop.
Then I turned off the archive Wi-Fi and used mobile data.
The exhibition page loaded.
I opened several images, checked their captions and followed the attribution links back to Commons. The collection was public, the files displayed correctly and the links printed on the gallery labels now led somewhere useful.
That second device had not solved the restricted login. The laptop connection had already done that.
It removed the final uncertainty. I could verify the public page through a separate network without beginning another registration or account-recovery process.
At 11:52, the archive director entered the room carrying a box of labels.
“Can we print?”
I turned the phone toward her.
“Yes.”
Indonesia made the missing step more important than the visible site
The Wikimedia restriction was later lifted after the foundation completed its registration. On April 30, the ministry announced that access was being normalised.3
The incident still revealed an important distinction.
A website can appear available in Indonesia while its login, upload system or another essential dependency remains unreachable. From the outside, the service looks open. The task the user came to complete is still impossible.
The established provider had the longer record, broader network and larger body of independent reviews. The smaller service’s shorter public history remained its clearest limitation.
But the archive did not need hundreds of possible destinations.
It needed one route through the network to the blocked authentication page.
The large provider gave me countries and protocols to test. The smaller app treated the restrictive connection itself as the problem, opened the missing page and let the archive reach the public before its doors opened.
For that morning in Yogyakarta, the best VPN for Indonesia was not the one that could load Wikipedia. It was the one that could reach the part of Wikipedia Indonesia had made invisible.
Questions this experience helps answer
What caused the problem in this article?
That difference had seemed minor until I was standing between a blocked login and a folder of large files.
Why did the obvious first fix fail?
The visible difference was clear: the established provider could not hold the route open, while the smaller app connected over the same Wi-Fi and carried every image to Commons.
What changed when the task finally worked?
The task worked when the connection matched the real workflow and remained usable through the important step, rather than merely showing a connected status or a fast local speed test.
What should someone check first in a similar situation?
I could verify the public page through a separate network without beginning another registration or account-recovery process.