My '99.9% uptime' provider had 47 hours offline this quarter

So guys I signed up with Hetzner back in january, thier page says 99.9% uptime SLA... I run a small reseller operation, nothing fancy, just WHMCS and some shared accounts...

I kept track in UptimeRobot and got 47 hours downtime this quarter alone... did the math, thats like 94.8% uptime?? I opened a ticket and they gave me some canned response about "network maintenance windows not counting"

But like... 47 hours is almost two full days... my clients are asking for refunds... anyone else with Hetzner seeing this or did I just get unlucky

Any leads on providers who actually honor thier SLA? Looking at 2-3x price range is fine if they mean it... margins are thin but churn is worse

Fio Disk Speed Tests (Mixed R/W 50/50):
---------------------------------
Block Size | 4k            (IOPS) | 64k           (IOPS)
  ------   | ---            ----  | ----           ----
Read       | 14.23 MB/s   (3.5k) | 89.41 MB/s   (1.3k)
Write      | 14.25 MB/s   (3.5k) | 89.89 MB/s   (1.4k)
Total      | 28.48 MB/s   (7.1k) | 179.30 MB/s  (2.8k)

iperf3 Network Speed Tests:
---------------------------------
Provider        | Location           | Send       | Receive
                |                    |            |
Clouvider       | London, UK         | 892 Mbits  | 741 Mbits
Online.net      | Paris, FR          | 12 Mbits   | 8 Mbits
WorldStream     | The Netherlands    | 456 Mbits  | 398 Mbits

Decent disk IO, meh network

Hetzner London vs OVHcloud London last month:

• Hetzner: 3 outages, 47h total, avg 892 Mbps when up
• OVHcloud: 0 outages, 99.97% uptime, avg 341 Mbps

Your math checks out... 47h/quarter = 94.78% uptime

Prices only go up

SLA credits are a joke anyway... they make you beg for 5% of monthly... /24 will cost you a kidney by 2027 and these hosts still oversubscribe transit like its 2019

Oh no... thats not just unlucky... thats a pattern... I had similar with RackNerd last year... they counted "scheduled" after the fact... you check thier status page and suddenly everything was planned... even when you got no email...... five hour outage becomes "planned maintenance" with zero notice...... the SLA is basically a marketing number...... nobody reads the exclusions until its too late...

I moved to KnownHost and so far so good... but who knows... they all look the same until they dont......

Thank you very much sir

I made VPS with Hetzner last month... no issue yet but I am worry now... sir @RAJ you are check timezone in UptimeRobot? I am set wrong timezone before and think server down but actually was maintenance window in my night... sorry for bad english sir

I believe Hetzner's SLA excludes scheduled maintenance in UTC+8. Their status page shows 6 maintenance windows this quarter, total 44 hours. If your monitoring is in IST or similar, the overlap may appear as unplanned outage.

I checked their terms once for a Singapore client. The math changes significantly if 41 of 47 hours were announced.

¡oh no! I am very sorry to read this... I had similar problem with GreenCloudVPS and is very frustrating... I write long post about SLA deception but then I delete... Jajaja... better short version...

Vut you are check the timezone??? I am make same mistake before... my "downtime" was actually they restart at 3am local... I feel so stupid when I see... the 47 hours maybe is 6 hours real??? ¿you verify this?

I hope is not so bad... good luck amigo

Same at my old host, network maintenance never counted either

Network maintenance windows don't count? That's half the quarter