1 |
Am Wed, 21 Sep 2016 13:47:28 -0700 |
2 |
schrieb Grant <emailgrant@×××××.com>: |
3 |
|
4 |
> >> I haven't mentioned it yet, but several times I've seen the website |
5 |
> >> perform fine all day until I browse to it myself and then all of a |
6 |
> >> sudden it's super slow for me and my third-party monitor. WTF??? |
7 |
> > |
8 |
> > I had a similar problems once when routing through a IPsec VPN |
9 |
> > tunnnel. I needed to reduce MTU in front of the tunnel to make it |
10 |
> > work correctly. But I think your problem is different. |
11 |
> |
12 |
> |
13 |
> I'm not using IPsec or a VPN. |
14 |
> |
15 |
> |
16 |
> > Does the http server backlog on the other side? Do you have |
17 |
> > performance graphs for other parts of the system to see them in |
18 |
> > relation? Maybe some router on the path doesn't work as expected. |
19 |
> |
20 |
> |
21 |
> I've attached a graph of http response time, CPU usage, and TCP |
22 |
> queueing over the past week. It seems clear from watching top, iotop, |
23 |
> and free than my CPU is always the bottleneck on my server. |
24 |
|
25 |
What kind of application stack is running in the http server? CPU is a |
26 |
bottleneck you cannot always circumvent by throwing more CPUs at the |
27 |
problem. Maybe that stack needs tuning... |
28 |
|
29 |
At the point when requests start queuing up in the http server, the load |
30 |
on the server will exponentially rise. It's like a traffic jam on a |
31 |
multi lane high way. If one car brakes, thinks may still work. If a car |
32 |
in every lane brakes, you suddenly have a huge traffic jam backlogging |
33 |
a few miles. And it takes time to recover from that. You need to solve |
34 |
the cause for "braking" in the first place and add some alternative |
35 |
routes for "cars that never brake" (static files and cacheable |
36 |
content). Each lane corresponds to one CPU. Adding just more lanes when |
37 |
you have just 4 CPUs will only make the lanes slower. The key is to |
38 |
drastically lower the response times which are much too high if I look |
39 |
at your graphs. What do memory and IO say? |
40 |
|
41 |
-- |
42 |
Regards, |
43 |
Kai |
44 |
|
45 |
Replies to list-only preferred. |