0

I'm having a problem with HAProxy. We have a load-balancer running haproxy with SSL offloading, which forwards traffic to our several app servers, running Windows Server 2008 and IIS 7.5. The problem I'm having is that when I deployed the latest version of one of our apps, HAProxy ceased to forward traffic to any of the app servers for that subdomain. Instead, we saw instantaneous HTTP 503 errors.

I've confirmed from IIS logs that no traffic reaches the server. I've attempted to read through the haproxy logs, but I'm woefully unfamiliar with the app and most of the log entries seemed to address various flavors of memory allocation calls.

EDIT

It seems that I'd posted the question without finishing the text. The sentence previously ending in "flavoe" has been corrected.

2
  • 1
    What does the haproxy status page say? Also, just because you can't understand the logs doesn't mean that nobody else can understand the logs. Also, what is a "flavoe"?
    – womble
    Aug 4, 2011 at 2:07
  • Don't shoot the messenger. If you haven't changed anything in haproxy, and it was working before, then it means the problem is most likely in your app or IIS server configuration. Aug 4, 2011 at 2:26

2 Answers 2

0

I would start with something simple like: Is IIS running on the same port after the deploy as before? Hard to forward if the port has changed.

0

As was perfectly expected, the problem was with the application code. An error caused a certain page to return an HTTP 500 error. Unfortunately, that was the URL that haproxy used to monitor uptime. As soon as the new code was deployed, haproxy was informed that the site was unavailable.

The answer was to correct the application code, obviously. What's not obvious is how to determine that this is what was happening - I ended up resorting to tcpdump to diagnose traffic, as haproxy logs were less than informative.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .