Extrememly slow performance every 30-60mins over the last 3-4 days

Website, Application, Performance
1

What is the name of the domain?

https://i.mjh.nz/SamsungTVPlus/.channels.json

What is the error number?

No error. Just eventually times out

What is the error message?

No error. Just eventually times out

What is the issue you’re encountering

Extrememly slow performance every 30-60mins over the last 3-4 days

What steps have you taken to resolve the issue?

Turned off all security settings.
Tested with files not via Cache (still has same issue)
Turned off http/3. Turned off speed brain beta

What are the steps to reproduce the issue?

Request https://i.mjh.nz/SamsungTVPlus/.channels.json every 60seconds.
A few times per hour, it will slow down to snail pace.
Then comes back to normal.

Users downloading it from various different places around the world have same issues.

I disabled cache on an identical file and it has the same issues at the same time.

I have another JSON file on same server that the issue does not happen with: https://i.mjh.nz/frndly_tv/app.json

Screenshot of the error

timing.png
timing.png1697×399 7.89 KB

4 Likes
2

More screenshots:

uptime
uptime840×921 30.8 KB

3 Likes
3

and another. this is the response headers:

headers
headers2547×572 21.6 KB
i

2 Likes
4

one more of uptime over last 6 hours for one of the files

Screenshot 2024-11-13 104607
Screenshot 2024-11-13 1046071627×805 66.5 KB

3 Likes
5

Note: The .xml.gz in the uptime screenshots are actually hosted on github so CF is just redirecting. That is why they are not affected

All the rest live on the same origin.
So its interesting does not happen with the frndly_tv/app.json
That would be smaller than these other json files - so maybe the size is an issue?

Also, again - tried with an identical file set to bypass cache and same issues at same time.

3 Likes
6

+1, I’m curious to know what the problem is here, it’s causing downstream problems

7

+1 This issue impacts downstream applications that depend upon what @Matt_Huisman has highlighted, causing the need to create time and resource intensive workarounds.

8

the issue is spreading to other files now:
eg. https://pinn.mjh.nz/data/app.json

9

another interesting find.
The downtime is almost always 10mins

image
image1171×819 24.7 KB

10

this is still an issue across multiple domains…
really dissapointing :frowning:

1 Like
11

I created a new subdomain today and turned off CF proxy for it (gray cloud)
Everything else is identical. nginx updated to listen on both domains.
No issues on the requests going directly

1 Like
12

compare
compare1577×1100 38.2 KB

13

Do you see any origin errors (codes 520 and above) when you look at the logs on dash.cloudflare.com?

14

im on a free acount, not sure where to find those logs

I was using the legacy page rule with a single rule matching all to cache everything

I have just deleted that and swapped to the new “Cache rules” instead.

Will see if that helps…

15

i do notice using the new cache rules, the headers have changed slightly

image
image1667×363 14.3 KB

So, fingers crossed…

16

unfortunately that didn’t fix the problem. Getting slow performance again now

17

bump… I’ve tried as much as I can.

Its quite obviously a CF issue.
Im just not sure if its intentional or not and would like some clarity

Next thing I may try is using a CF tunnel instead of “proxy”

18

You can’t use a tunnel without the proxy, afaik.

19

yup, but maybe it goes via some different pipeline that wont have the same issue?

20

Couldn’t hurt to try.