New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Anyone with a web server on Ramnode's OpenVZ?
Mine's a 512MB OpenVZ running nginx and when I upload a test html file
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Test</title>
</head>
<body>
Test html file
</body>
</html>
and rush test that html in blitz.io (just the free test), I always get an error and or timeout. I cannot adjust the sysctl.conf and I'm guessing that's where the tcp timeouts and errors are coming from? This means if you have a popular website, even if it's just static html, some of your visitors may see an error or timeout.
Could anyone try this too on their OpenVZ VPSes? Blitz.io offers 10 free credits per month. 1 credit for each test.
On the other hand, I did the same test on a KVM VPS with adjusted sysctl.conf, no errors or timeouts.
Comments
I saw the same thing on my NYC box, but I had a hell of a lot more timeouts than you did.
25,615 hits with 52 errors & 1,651 timeouts.
https://oc.charlesauer.net/public.php?service=files&t=1037afd0aed9acfdd31db7c660c72f1c
Not sure if the timeouts are due to CSF or not.
I would guess that your nginx config can't keep up with the 500+ req/sec you have going there. Have you optimized your nginx config?
I find this guide really useful http://blog.zachorr.com/nginx-setup/
wow same here. My OpenVZ is also in NYC. I wonder how's the performance of the other datacenters (NL, Atlanta, Seattle) in the same test.
@stoke the KVM (not Ramnode) was able to handle it. same nginx.conf
I haven't really messed with the base nginx config much, so it's pretty much the default, but I've got varnish running in front of nginx and it reported no errors during the test.
What about socket exhaustion? How many sockets are open/available?
If I'm reading netstat right, I've got 8 for nginx, none for varnish.
Vps in Atlanta
VZ shares connections on host's kernel. It's quite normal to see timeout if sockets are all used by your neighbors.
So basically i just need some OpenVZ containers on the same node, with much request per Hour and my Page can suffer? Even if i only get 5 request per hour interesting. One point more to move away from OpenVZ. What about XEN-PV?
KVM FTW.....also use ramnode promo code to bring the price down
On the bright side, I got rid of the timeouts that varnish was throwing (but still leaving some errors), by adding the headers from the last post in this thread: https://www.varnish-cache.org/forum/topic/453