I have been encountering:

An error has occurred - please retry your request

Looking in the Chrome console when this happens, I see:

status 503 (backend read error)

That particular example occurred on Database Administrators when voting to delete a comment. I have also had it when voting on questions and answers.

I have had these red box errors in the past on the odd occasion, but it seems to be much more frequent over the past day or two. I've only recently learned about the Chrome console (thanks Shog9 & ArtOfCode), so I can't say if past errors would have had the same detail.

While editing this question to respond to a comment, I received:

An error occurred submitting the edit


status 503 (backend read error)

So it seems not to be limited to the site where I am a moderator. Logging out of Stack Exchange everywhere and clearing all my sessions and cookies did not resolve the problem.

Latest example:

An error occurred while loading the review item

status 503 (backend read error)

  • Have you tried logging out and back in? – nicael Jun 15 '16 at 7:17
  • hmm... so weird. Let's hope a developer will take a look soon and fix it. :) – Shadow The Princess Wizard Jun 15 '16 at 7:17
  • 3
    @nicael A physicist, a chemist, and a programmer were in a car that wouldn't start. The physicist examined the spark plugs and the engine; the chemist assessed the chemical makeup of the diesel and the oil; the programmer got out of the car and then went back in again. – Rand al'Thor Jun 15 '16 at 10:29
  • @randal' lol :D – nicael Jun 15 '16 at 10:30
  • looks like a network blip – Sklivvz Jun 15 '16 at 11:44
  • 3
    @Sklivvz: except that SO recently switched CDN proxy providers and we've been seeing all manner of issues around that. Are you sure it is a network blip and not another timeout race? – Martijn Pieters Jun 15 '16 at 12:15
  • 1
    We've added some indexing that should help with this. Can you let us know if this is still happening? – Oded Jun 15 '16 at 12:54
  • Ugh. And nothing in the exception logs :( – Oded Jun 16 '16 at 7:30

I originally thought it was an internal network blip between our load balancers and the web servers. It turns out that another explanation is likely correct.

First of all, we've upgraded SQL Server to 2016 last week. This is great for a number of reasons, but we (and by "we" I mostly mean Nick Craver) have been chasing performance issues since then.

Secondly, having switched to fastly, slow performance might be reported to users as 503, whereas we don't see corresponding exceptions in the logs.

We're on it.

You must log in to answer this question.