Skip to content

One year uptime on my Hostinger $10/year box, here's what I learned

VPS Hosting by lee_mcr 26 replies 2.6K views
#11

I'm on OVH for my cheap VPS, 3.50 EUR/month at Gravelines. Uptime's fine but the network... sometimes it's a lottery. Hostinger at 10 dollars a year is insane pricing, even with the asterisks.

@lee_mcr did you ever test their tech support? Because at that price, OVH's support is "read the documentation".

Vive la résistance... électrique
#12
lucgone said:
Because at that price, OVH's support is "read the documentation".

Once, early on, because I locked myself out of SSH with a bad `sshd_config`. Ticket answered in 4 hours with a link to their KVM console, which worked. Not amazing, not terrible. For 10 dollars I would have accepted "reinstall and restore from backup" as reasonable.

OVH at 3.50 EUR is competitive though. Their network lottery is real, heard the same from a mate in Roubaix.

Honey badger don't care... about downtime
#13

OVH Gravelines to Lisbon is consistently 35ms for me, but I pay for their VPS Comfort line, not the cheap tier. The cheap tier routes through Paris in ways that make no topological sense. You get what you pay for in routing table sanity.

@lee_mcr Hostinger's Vilnius datacenter: what's your latency to Manchester? I'm curious if they're using anycast or just straight georouting.

#14
netop_mares said:
Hostinger's Vilnius datacenter: what's your latency to Manchester

I'm on their London DC actually, not Vilnius. 12-15ms from Manchester. Vilnius would be more like 45-50ms which is why I picked London during signup. They let you choose at purchase, though it's buried in the flow.

Your point about OVH routing is spot on. I've seen traceroutes from their budget tier that look like modern art.

Honey badger don't care... about downtime
#15

¡Ah! I forget to say,,,,, My Hetzner was in Falkenstein,,,,, From Buenos Aires is 220ms always,,,,, But stable 220ms,,,,, Like a rock,,,,, A very slow rock

Now I am in Oracle Cloud free tier,,,,, Is "always free" they say,,,,, We will see,,,,, But Sao Paulo region is 35ms for me,,,,, Much better for VPN

@lee_mcr You have problem with DMCA or abuse complaints in Hostinger? In Hetzner I had one for "scanning" that was just my fail2ban being too loud,,,,, Very annoying!!!

#16
pablowild said:
You have problem with DMCA or abuse complaints in Hostinger

None, but my use case is boring: personal DNS, a static site, a private Matrix server with three users. The most exciting thing I host is a weather API scraper that hits meteoblue twice a day. fail2ban logs are quiet, no port scanning, no torrents.

Oracle free tier is a known quantity now, been around long enough that people trust it won't vanish. The ARM instances are genuinely good.

Honey badger don't care... about downtime
#17

Oracle free tier ARM is decent but the "always free" has limits they don't advertise well. I hit the 200GB boot volume cap and had to migrate. Their Tokyo region is oversubscribed during market hours, latency spikes to 200ms+ on the VCN internal routing. Still free though. Hard to complain.

@lee_mcr what monitoring do you use beyond UptimeRobot? Self-hosted or other SaaS?

8-char NTLM found in 4min 32sec
#18
Hana5 said:
What monitoring do you use beyond UptimeRobot

Just UptimeRobot free for external, and `netdata` on the box itself for internal metrics. Netdata is overkill for 512MB but I disabled most plugins, only keeping CPU, memory, disk, network. Runs fine.

I had Prometheus + Grafana on a separate box but that died to OOM during a Docker update and I never rebuilt it. The irony of my monitoring stack needing more resources than what it monitored was not lost on me.

Honey badger don't care... about downtime
#19
lee_mcr said:
The irony of my monitoring stack needing more resources than what it monitored

This is why I run my monitoring on a separate cheap box. Currently a Hetzner CX11 in Helsinki — https://www.hetzner.com/cloud, 2.76 EUR/month, runs Uptime Kuma, Prometheus, the works. My main boxes can die without taking observability with them.

The real cost is the time to maintain it. I spend more hours on the monitoring box than anything it watches.

...
7 #20

Uptime Kuma is acceptable. I prefer simple: `monit` on each host, status to a central `rsyslog`. No web interface, no JavaScript, just text files and `grep`. If I need a graph I use `gnuplot` like a civilized person.

@quinnbizz Uptime Kuma in Docker? Or bare metal?

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft