TierHive - 768mb ram | 7 GB NVMe | $0.64 /mo | UK, DE, USA, CA, SG, FR | Free to try| Build a plan!

AnthonySmith

@WSS said:
I checked the FAQ and see that only 3 TCP ports are forwarded per VPS. I can't find a way to edit them, though. How do I set the forward?

On the VPS management page in the server information container, top right, add more ports

WSS

Oh. There.

I would have expected it under the IPv4 config page. It makes sense to be able to edit there since it shows the allocation on that page.

AnthonySmith

@WSS said:
Oh. There.

I would have expected it under the IPv4 config page. It makes sense to be able to edit there since it shows the allocation on that page.

yeah, tbh i started thinking that when answering you, will make some notes, this is the result of changing direction half way through, initially haproxy and the network config and vps management were all 1 page... lets just say it got a bit... silly.

WSS

@AnthonySmith said:

@WSS said:
Oh. There.

I would have expected it under the IPv4 config page. It makes sense to be able to edit there since it shows the allocation on that page.

yeah, tbh i started thinking that when answering you, will make some notes, this is the result of changing direction half way through, initially haproxy and the network config and vps management were all 1 page... lets just say it got a bit... silly.

Yeah the flows are either really smooth, or "Who the fuck who thought a 10x10 mystery meat graphic was acceptable in 2026."

WSS

For what it's worth, given the resource and expense, it's pretty much just what I'd expect. It's a lot slower than other KVMs, but I don't really expect much for under $0.50/mo. The HAProxy config was very simple, even though having it throw 503s at me for minutes because I was throwing 404s was confusing.

WSS

Also, along with the 503s (capacity?) HAProxy isn't passing the client IP address, so my simple rate limiting script is unhappy with a couple of hits to my contact page.

AnthonySmith

@WSS said:
Also, along with the 503s (capacity?) HAProxy isn't passing the client IP address, so my simple rate limiting script is unhappy with a couple of hits to my contact page.

503 is saying it can't reach your back end server

If you end up 404 instead of 200 then it kicks your record out, it can take 15 minutes for it to be back in once it's responding with 200's again

WSS

It was doing fine with a local netcat. (shrug) It seems to be kind of sporadic. I've got a pre-cached nginx and grav with a handful of php-fpm in the pool. Even if php-fpm went down for several hours, it should still serve from my nginx cache.

Only thing that it might trip would be the too-many-consecutive hits pool, and that's only set for the contact page.

AnthonySmith

@WSS said:
It was doing fine with a local netcat. (shrug) It seems to be kind of sporadic. I've got a pre-cached nginx and grav with a handful of php-fpm in the pool. Even if php-fpm went down for several hours, it should still serve from my nginx cache.

Only thing that it might trip would be the too-many-consecutive hits pool, and that's only set for the contact page.

I will have a look in a bit, if it's our side being over the top I will relax it

WSS

@AnthonySmith said:

@WSS said:
It was doing fine with a local netcat. (shrug) It seems to be kind of sporadic. I've got a pre-cached nginx and grav with a handful of php-fpm in the pool. Even if php-fpm went down for several hours, it should still serve from my nginx cache.

Only thing that it might trip would be the too-many-consecutive hits pool, and that's only set for the contact page.

I will have a look in a bit, if it's our side being over the top I will relax it

I've been watching your curl hits every so often and they're fine. They do hit my default page, though, and not the one I've got configured. You're doing an HTTP/1.1 request, but I don't see my hostname there. My default failthrough IP based auth/unknown doesn't (didn't) have an index page so it was throwing a 404 at you.
%sudo tail /var/log/nginx/access.log
10.0.0.15 - - [22/Apr/2026:11:58:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:00:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:02:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:04:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:06:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:08:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:10:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:12:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:14:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:16:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"

Compared to other traffic-

%wc -l /var/log/nginx/site.com.access.log
254 /var/log/nginx/site.com.access.log
%

AnthonySmith

@WSS said:

@AnthonySmith said:

@WSS said:
It was doing fine with a local netcat. (shrug) It seems to be kind of sporadic. I've got a pre-cached nginx and grav with a handful of php-fpm in the pool. Even if php-fpm went down for several hours, it should still serve from my nginx cache.

Only thing that it might trip would be the too-many-consecutive hits pool, and that's only set for the contact page.

I will have a look in a bit, if it's our side being over the top I will relax it

I've been watching your curl hits every so often and they're fine. They do hit my default page, though, and not the one I've got configured. You're doing an HTTP/1.1 request, but I don't see my hostname there. My default failthrough IP based auth/unknown doesn't (didn't) have an index page so it was throwing a 404 at you.
%sudo tail /var/log/nginx/access.log
10.0.0.15 - - [22/Apr/2026:11:58:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:00:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:02:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:04:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:06:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:08:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:10:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:12:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:14:01 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"
10.0.0.15 - - [22/Apr/2026:12:16:02 -0500] "GET / HTTP/1.1" 200 615 "-" "curl/8.14.1"

Compared to other traffic-

%wc -l /var/log/nginx/site.com.access.log
254 /var/log/nginx/site.com.access.log
%

Right, that's a pretty obvious problem,can't believe no one else has reported it.

I will have some changes made.

WSS

@AnthonySmith said:
Right, that's a pretty obvious problem,can't believe no one else has reported it.

Probably because you're throwing a 503, which looks like misconfiguration or a capacity problem. That's what confused the hell out of me, until I saw "Wait, this heartbeat isn't following spec."

WSS

Is there anyway to resize down from 10GB? It won't let me, and I'm using maybe 4. Unless it's a qcow image, that's just waste.

AnthonySmith

@WSS said:
Is there anyway to resize down from 10GB? It won't let me, and I'm using maybe 4. Unless it's a qcow image, that's just waste.

It's qcow and it's based on the distro official image minimums, or maybe you selected 10gb on create.

Downgrading disk is not supported (yet) but I might be able to do it manually for you if you want

AnthonySmith

@WSS said:

@AnthonySmith said:
Right, that's a pretty obvious problem,can't believe no one else has reported it.

Probably because you're throwing a 503, which looks like misconfiguration or a capacity problem. That's what confused the hell out of me, until I saw "Wait, this heartbeat isn't following spec."

You now get about 20 minutes of local downtime before haproxy boots your record out instead of 4 minutes.

We are also going to talk about a generic enough but informative enough 503 so you know where its being stopped should it be hit by haproxy.

@WSS said:
Is there anyway to resize down from 10GB? It won't let me, and I'm using maybe 4. Unless it's a qcow image, that's just waste.

Token adjustment for that vps applied so you are only paying for what you are using.