Getting ocassional 503 errors on azure website

Question

I'm getting occasional 503 errors on our site. It usually happens after not visiting the site for a while. The whole page might return 503 or just some resources like css or js files.

It seems to go away after you've surfed the site for a bit and hit all of our servers.

Elmah doesn't show any errors.

I've gone into the logs on each of our servers (three medium web roles on azure) and I can't find any problems.

Our deployment has been up since December without a code change, we've been having this problem for about a week.

One thing to note is that when this happens the site doesn't shut down. I would think that would happen if IIS was crashing and restarting (even with three servers).

Does anyone know how to diagnose or fix this problem?

Answer 1

While this could be code related, I'll assume you've already explored this route as much as possible via logs (and since you haven't deployed new code). Having said that:

Do your issues align with the Compute service degradation events shown in the Azure Dashboard over the past several days? Look at Historical View and you'll see a few issues around Compute. Depending on your data center, maybe this is related?