from Hacker News

Cloudflare is experiencing elevated 5xx responses

by danielandrews43 on 12/15/23, 10:06 PM with 41 comments

• by sillysaurusx on 12/15/23, 11:34 PM

  Apropos nothing, I found out today that bash.org died. https://news.ycombinator.com/item?id=37295238 was the last I was able to find about it, and it's the top google search result.

  I've been thinking of the best way to make a website that won't die when I die. It seems like a hard problem. I was originally going to just trust that Github Pages will be around forever, but I'm not so sure anymore.

  Running your own server is fraught with danger. In theory it's just a matter of making sure your account has enough credits to last 100 years. In practice, I can't count the number of times that servers have had one problem or another that requires manual intervention to resurrect.

  It's surprising that this isn't a solved problem. "Put some HTML somewhere" has always left the question of "where?"

  Maybe paying for your own S3 account and shoving the HTML onto that is the way to go. Still, thinking of the number of companies that have survived even 20 years, the odds aren't good that both S3 and Cloudflare will run flawlessly for decades.

  Oh yeah, the reason we're able to read these comments now is because Dan migrated HN away from Cloudflare, I think to Route 53. So maybe S3 + Route 53 has the highest odds of standing the test of time.

  Then there's the question of what to write. But at least "test website please ignore" has a chance of lasting a century, unlike everyone else.

• by spindle on 12/15/23, 10:54 PM

  I keep imagining an elevated 5xx response is a 599 when it should be a 503 or a 504.

• by lemoncookiechip on 12/15/23, 11:53 PM

  Discord is having issues right now, unsure if it's related.

  - Investigating - We are working to correct an issue that is causing some servers to be degraded and unavailable. - Update - We are continuing to investigate this issue.

• by ciceryadam on 12/15/23, 10:48 PM

  It's fixed, at least for us.

  We had a bunch of 530 codes returned on our endpoints, although Cloudflare says that they implemented a fix at 22:39 UTC, we don't have an increased rate of errors since 22:10 UTC.

• by devaiops9001 on 12/15/23, 11:18 PM

  Are databases of Cloudflare's core services deployed in more than one AZ yet?

• by tremon on 12/15/23, 10:57 PM

  Ah, that might explain why I was seeing http 520 errors from various sites. It was the first time in a long time that I needed to lookup a http response code, only to find it's an unofficial one.

• by Ocha on 12/15/23, 11:42 PM

  Didn’t experience any outage for mailsnag.com. We have newrelic synthetic checks around the globe, and another uptime monitor. Not even a blip on the charts.

• by RyanGWU82 on 12/15/23, 10:22 PM

  I'm also seeing a lot of 1016/1018 "Origin DNS errors" across a variety of web services. I saw this on my own company's website first around 21:40 UTC, although it's not bad enough to materially affect our traffic volume.

• by ChrisArchitect on 12/15/23, 11:31 PM

  Tell HN: