Closed Silver-Golden closed 4 years ago
Hi @Silver-Golden ,
It seems that it is a little hard to find the root cause. Maybe we should try more tests to dig it out. May I know how your Caddyfile looks like?
Hi @Silver-Golden ,
The following image is the health check for the cache server (which is built on version v0.3.8)
I don't encounter the situation you mentioned. There are just fewer picks because of the latency from far region.
In addition, I've also registered the monitor service (which you used)
This also looks good.
I think the problem is not related to the cache server.
I completely forgot to follow this up, I was busy with a rust project and lost track of time.
I'll dig in a bit deeper today, to see if I can recreate it and if I can when the problem arose.
@Silver-Golden ,
Never mind, I just curious about what happened. BTW, I think the uptimerobot's health check is not stable. I will show you what I find. Go to see the following two images.
it shows a high pick below.
However, there is not such high pick in the google's uptime check.
PS: In order to compare, both of the time zones are set on UTC+8.
So if you configure the timeout(maybe 1 second) in the uptimerobot, it will judge the cache server is down when the peak appears.
Maybe, you can use another solution to monitor the cache server. Ex. google's uptime checks. It's so steady :) (you can see that from the above image).
So I used 0.3.7 to compile it.
I then waited about an hour before enabling teh cache for this endpoint.
When I did enable it the periodic timeout and access started happening.
xcaddy build v2.1.0-beta.1 --with github.com/caddy-dns/cloudflare --with github.com/sillygod/cdp-cache@v0.3.7
I'll do 0.3.6 next
Update
Same procedure as above:
The red sections in order are 0.3.7 (as above), 0.3.6, 0.3.5 and 0.3.4 which has no red/timeouts.
I suspect that something changed between 0.3.4 and 0.3.5 that is giving this behaviour.
I am still not sure if its actually impacting users but it is spamming my inbox
Ugh as soon as I posted that I refreshed the monitor and seems like 0.3.4 is giving a timeout now, back to testing
0.3.3 gaves timeouts as well and now back to 0.3.2 which is not giving the timeouts.
Looking back I probally should have started with 0.3.2 and gone up
I consider closing this issue if there is no example to reproduce this bug. Until now, I still can not find a way to reproduce this one.
I had an issue earlier.
I use uptimerobot to monitor my uptime and to notify me if stuff goes fubar.
Today I had this.
(Sorry the numbers are a tad small, my paint skills arent the best)
http_cache
was not a directive.xcaddy build v2.1.0-beta.1 --with github.com/caddy-dns/cloudflare --with github.com/sillygod/cdp-cache
xcaddy build v2.1.0-beta.1 --with github.com/caddy-dns/cloudflare --with github.com/sillygod/cdp-cache@v0.3.2
Because compiling to v0.3.2 worked and v0.3.8 didnt I presume something between the two is causing issues.
Now that I am awake again I will do some further testing to hone in on which version introduced the issue.
Logs