1
votes

We are hosting several ASP.NET MVC sites on Azure (Wesites in standard mode) and recently we have run into an issue with high latency - the server takes very long time to respond even for static files.

The developer console in Chrome shows these data for a sample image file (100kB):

  • connecting 12ms
  • sending 3ms
  • waiting 1856ms
  • receiving 89ms

We have configured autoscaling, so the website can scale out if the server gets to busy, but it doesn't seem to help.

Any idea, where the problem could be?

Edit:

The site is hosted in the West Europe region.

It appears, that the problem isn't in our ASP.NET application. I have created a new website that contains only static files and response times are also bad.

2
I don't see anything on the service dashboard (windowsazure.com/en-us/support/service-dashboard). How recently have you been encountering the issue and for what duration?Greg D
Are you still experiencing the problem, I am not able to replicate it with my AzureWebsites. Can you share the URL with which you are encountering the problem? Is this issue happening in a timely manner, I mean daily at some times? Do you see any unusual activity on Website Monitoring Dashboard?ramiramilu
It started yesterday and since then, the performance is consistently bad. There isn't anything unusual on the monitoring dashboard. URL of the website is live-zsvhejny-dmp.azurewebsites.netLukas Kabrt
I do see a similar degradation of response times with our WAWS. Standard mode in region West Europe.twomm
What is in the eventlog.xml file?Alexandr Nikitin

2 Answers

2
votes

Switching to instance size M has solved this for us ... for now.

0
votes

I copy-paste the answer from my question because I think we have the same problem

I got a reply from Windows Azure Web Sites Team on MSDN forum:

We have identified this issue as a bug in new Auto-Heal feature we shipped recently. Please add following in your web.config file to mitigate this issue for now.

<configuration>
    <system.webServer>
        <monitoring>
            <triggers>
                <slowRequests timeTaken="02:00:00" count="1000000000" timeInterval="00:01:00" />
            </triggers>
            <actions value="LogEvent" />
        </monitoring>
    </system.webServer>
</configuration>

We are working on releasing the fix ASAP.

Apologies for inconvenience.

Windows Azure Web Sites Team

Update: The workaround works only in Standard mode. It throws an error in Shared or Free mode, so be aware.