Due to extreme load, this is temporarily being shown to everyone as a logged out user would see it

support
1

We had a surge of new visitors this morning due to a popular AMA and many people received this message which was somewhat understandable since one topic was getting overloaded and the cpu spiked during this time. Since then it has subsided but I keep getting this error telling me I’m logged out then it refreshes and shows the warning message.

FYI I have 2gb memory and 1 cpu on DO. There were about 30 users online this morning when it first happened and now there are only 10. Do I just need more CPU power or does this seem a bit odd that it keepshappening?

image
image1964×1120 170 KB

2 Likes
2

How long has the warning gone on? Have you tried restarting the Discourse Docker container and/or server?

3 Likes
3

It’s been off and on, I have seen it four times today but most of the time it works fine. I just restarted docker using service docker restartis that correct? If it keeps happening I’ll try rebooting, trying to avoid the downtime since we have a lot of new users checking out the site. Thanks Jeff.

5 Likes
4

It should come back automatically, @sam built this feature and I know it has worked for us in the past as designed… maybe he has some ideas?

5 Likes
5

Even if there are only 10 users online a large number of users could be doing background work and an enormous of anon users could still be hitting it.

The site should auto recover as soon as the load subsides.

If you really would like to rush it you could try restarting the container.

5 Likes
12

I has same problem

image
image1463×871 427 KB

Use online 5

image
image1256×776 162 KB

My server

image
image733×785 75.9 KB

13

Disable all your third-party plugins, rebuild, and see if the problem persists. That’s the first step.

3 Likes
14

@davidkingham You use the browser?

15

I am having this message too. Could you please explain what condition triggers this message from Discourse? Is it when all Unicorn workers are busy? Some server metrics? Thanks.

16

Have you tried the steps Jeff and Sam mentioned to resolve it?

17

Will do, but I nevertheless would like to understand what is being monitored and would appreciate an answer to my question.

18

https://github.com/discourse/discourse/blob/ecb098bcbb0aad9e7d3777e036e4eafb0cfece9b/lib/middleware/request_tracker.rb#L155-L161

https://github.com/discourse/discourse/blob/ecb098bcbb0aad9e7d3777e036e4eafb0cfece9b/config/nginx.sample.conf#L100-L100

Amount of time it takes a request to find its way from NGINX to the Unicorn worker process.

3 Likes
19

Thanks, @sam, that is very helpful. Now I better understand why a plugin (which is making too many ajax calls) could be causing the problems we are seeing on our site. Besides the “extreme load” message, we are also seeing white “slow down” screens (too many requests, nginx), which would also fit into the picture.

We are now in the process to identify the culprit. One problem we already found is Babble. Our chat was well over 10k posts, which was causing heavy load on the system:

I’ll report back once we know more.

3 Likes