Forbidden City Night Festival Ticketing Chaos and How to Recover a Crashed Website

Today is the Lantern Festival, and many places are holding lantern events. The Palace Museum announced that on February 19 (the 15th day of the lunar new year) and February 20 it will host the "Night of the Forbidden City" cultural activity, marking the museum’s first night‑time opening in 94 years and offering free admission.

Free, hurry up!

People are already imagining strolling hand‑in‑hand through a romantic night tour of the Forbidden City.

However, tickets must be secured on the official website, and the difficulty of obtaining them is comparable to the intense competition for train tickets during the Spring Festival.

On February 17 around 14:40 the museum quietly opened the reservation system for the public; within an hour the tickets for February 19 were sold out.

Enthusiasts prepared to grab tickets for February 20 in the early hours of the 18th, but the website crashed before midnight, as shown by the screenshot of the failure.

After waiting for more than half an hour, users finally accessed the event page around 1 am, only to find that all three time slots had zero remaining spots, prompting a wave of public complaints.

"The Palace Museum reminds you: there are countless roads, but you have none." "I refreshed for an hour, feeling like smashing my mouse, only to be stuck." "Professional runners refreshed for an hour, saw zero spots, and hope for a future encounter."

One netizen lamented that without experiencing a server crash, you never understand how badly the public suffers, comparing it to the well‑known ticket‑grabbing issues of train ticket platforms and other popular sites.

Although the system recovered near midnight, the tickets were gone, and some scalpers began posting resale tickets on other platforms.

Why did the website crash, and how should it be quickly repaired?

Programmer Life (ID: coder_life) interviewed a senior operations engineer from Qunar.com to discuss website crashes. The engineer explained that the crash was caused by instantaneous traffic overload.

Traffic overload has two aspects: excessive concurrent requests that exceed the backend server’s capacity, and performance problems in the website code that slow down the system, leading to service failure.

There are two typical remediation methods:

1. To restore service quickly, limit access frequency dynamically, which ensures availability at the cost of restricting some users.

In normal circumstances, the maximum sustainable traffic should be determined through testing and capacity planning before the service goes live.

2. When time permits and the backend can be scaled, analyze the traffic data from the crash period, expand resources accordingly, and gradually lift the access restrictions.

After missing the night‑time festival, many suggested live‑streaming the event and adding more open sessions, hoping to keep traditional culture alive in the present.

Did you manage to get a ticket? What are your thoughts on ticket‑website crashes?