Uptime gives false positive.

My site is working fine, but am getting an email that it is down. Why is this?

  • Adam Czajczyk
    • Support Gorilla

    Hello Bamboo,

    I hope you're well today and thank you for your question!

    The way uptime check works (ours or pretty much any other) is that it pings the URL and waits for response. If the response doesn't get back in a set timeframe or there's too many packages lost on a way, check assumes that site's down and sends an alert.

    Basically, an alert from any uptime check service/tool shouldn't be considered as an ultimate "site is down" statement but rather a notification that "something might be wrong with site and it might not be accessible at the moment, please check it".

    With our check I think the timeout is around 20 seconds (though I might be wrong, it might be a bit shorter or longer) and if there's no response from the site within that timeframe, there's an alert. With other checks that timeframe might also be shorter or longer so results might vary slightly.

    The downtime that you check in cPanel, however, is a bit different thing and it mostly indicates that the server itself was running fine. That doesn't necessarily mean that site was accessible all the time. Furthermore, due to the way Internet works, if you setup two separate uptime check services for the same time, they also might bring different results. Why is that so?

    Because the "endpoints" would most likely be in different location. By "endpoint" I mean a checkup server that actually calls your site. That results in the fact that a request for site from such server and a response for that request would travel (at least partially) via different connections, which might provide different bandwidth or even be less or more "clogged" at the given moment when check is performed.

    That being said, I have checked your site multiple times with Pingdom tools and with just Chrome browser from my computer. The loading/response speed of your site vary greatly depending on the location from where the check was initiated. It might also vary a lot depending on the current load of your server. As a result, it might cause uptime check to be timed out while real user wouldn't notice site being down (maybe wouldn't even notice it being slightly slower for a minute or two), which in turn may cause such alerts.

    So, while they could be considered "false alarms", I would rather consider that a suggestion that it might be good to think of site performance optimization. The fastest load that I was able to get for the site while I was testing it, was nearly 8 seconds while average was over 10 seconds. That's a speed that affects user experience, might affect search results and suggests that sometimes that load may be longer, affecting uptime checks that way.

    I noticed that the site's hosted on SiteGround. Is it a shared hosting or is it some sort of VPS/Dedicated server?

    Could you also, please, enable support access to the site so I could take a look at the setup? You can do it by going to "WPMU DEV -> Support" page in site's back-end and clicking on "Grant Support Access" button there. I could then access the site, check it configuration and hopefully give you some tips on how to optimize/speed it up and/or what might be causing those uptime alerts.

    Also, I'd recommend implementing some CDN (Content Delivery Network) solution, like i.e. free CloudFlare, which may improve not only performance of the site but also make it more consistent across various locations (which is main point of CDN).

    Apart from this, we're constantly working on improving our uptime algorithms to minimize occasional "false alarms" as well, so I believe you'll experience less of them overtime anyway :slight_smile:

    Kind regards,

  • Predrag Dubajic
    • Support

    Hi Bamboo,

    Thanks for granting access.

    I was checking your sites with some additional tools like tools.keycdn.com/ping and geopeeker.com and I noticed that your ping times are really large and even the results are all positive in the end it takes a really long time to finish the load.

    Below is an example result which is similar to all of your sites:

    Singapore is the only location returning a good result, all the others are resulting in huge ping times.
    These times will affect your site load times and as Adam mentioned that can cause to show your site as down if it doesn't load it in predefined max time.

    Now, this might look like a false positive as the site is actually up but having these long loading times should be addressed in order to offer best experience for your visitors.

    I did check your site backend to see what kind of speed improvements you have active and the results are actually quite good, there's still some room for improvement by enabling Minification and Caching on your site.
    However it's still that ping time that needs to be resolved and this would be best discussed with your host and enabling CDN for your sites.

    Let us know if you have any additional questions about this.

    Best regards,

Thank NAME, for their help.

Let NAME know exactly why they deserved these points.

Gift a custom amount of points.