Twitter is up and down, with error messages and other outage-like problems

Twitter appears to be recovering after users spent hours on Wednesday evening reporting various errors, with the app and site throwing up errors about exceeded rate limits, failing to fetch new tweets, or appearing to be completely broken.

DownDetector.com showed a spike of around 10,000 outage reports noted for Twitter, with reports starting just after 7:30PM ET. By around 10:30PM ET, that number had dropped to less than 1,000 reports.

Here at The Verge, we had multiple instances of the desktop site logging accounts out suddenly. When one staffer clicked the log in button, the site would simply refresh without any prompt for a username or password — or crash, showing an error message. They have since been able to log back into their account and resume using the site as normal.

The mobile app worked for some people during the incident, though that wasn’t the case for at least one Verge staff member. The problems were hard to pin down — they varied, disappeared, and recurred as you can see from the bounce in this DownDetector chart.

Many users have reported seeing a number of issues with the service since Elon Musk fired a massive chunk of the company’s staff earlier this year. However, contrary to some predictions, the site has mostly stayed up since, even during events like the World Cup or during Elon’s self-reported instances of unplugging “one of the more sensitive server racks.”

It was, however, an open question how fast the site would be able to recover after hitting a snag, as has happened many, many times throughout Twitter’s history. In terms of time to recovery, Wednesday’s weirdness was obviously not the big one that some people have forecast, and the recovery didn’t end up taking terribly long.

It’s unclear what caused the issues — the Twitter Support account hasn’t mentioned the glitchiness at time of writing, though Musk tweeted that the company had rolled out “significant backend server architecture changes” to make the site feel faster at around midnight ET. It’s always possible the downtime was caused by the reconfiguration, but Musk’s only comment about the issues was a tweet saying that the site was working for him at around 8:30PM ET. The company did not respond to a request for comment from The Verge, but is widely reported to have disbanded its press office.

Update December 28th, 10:30PM ET: Updated to reflect that the partial outage appears to have ended.

Update December 29th, 12:45AM ET: Updated to add Musk’s comment about “significant backend server architecture changes.”