Has anyone successfully claimed SLA credits?

Wah this one worse than Vultr lor. At least Vultr whole region down then whole region down, no gaslight

My 5 provider setup: Vultr, Hetzner, Linode (now Akamai), DigitalOcean, and one mystery provider I won't name because too small. I run active-active with health checks. Last year total downtime from user perspective: 0 minute. Cost: high. Sanity: preserved. SLA credit collected: $0 because I never need to claim, just failover.

Not saying everyone should do this. But for company with revenue, engineering time > SLA fight time.

BuyVM? Frantech? I've been there. The real mystery is whether they'll exist next year.

I collect cheap VPS like stamps and the SLA question never even occurred to me. My Contabo box in St. Louis was $4.50/month when I got it. What credit am I going to fight for, forty cents? I just migrate my seedbox somewhere else when it dies. Currently running three concurrent Plex servers on three dead providers' annual plans I prepaid and forgot about.

This is why I negotiate upfront. My Hetzner dedicated was €89/month, got it to €76 with 24-month commit and no SLA credit clause. They wanted to keep the SLA marketing on website, I wanted actual reliability or actual discount. Win-win. Written into contract, not standard terms.

For cloud VPS this doesn't work. For dedicated or colo, always ask.

Luc never see SLA credit in France. OVH? Forget it. Their SLA is 99.9% but they count "planned maintenance" as not downtime even if announce 2 hour before. Scaleway similar. French providers have the "we are French, we do what we want" attitude.

I run monitoring from Lyon to my own box in Gravelines. When OVH burn, I know before their status page. Never claim, just accept and migrate. Life too short.

I made the spreadsheet. April: one incident, 8 minutes 100% loss, started 03:14 UTC. Status page: green. No maintenance posted. This should qualify under their terms.

But here's the thing. Their response time SLA is separate. Ticket #44291 took 65 days for first human response. That's also breach. I asked about both in escalation. The human only addressed availability, ignored response time entirely. Is this standard? Do they just not train support on response time SLA?

Also: I checked their status page history. Nothing for April 14. But I found a tweet from @ContaboStatus at 03:22 saying "network maintenance in progress." 8 minutes after my incident started. Deleted now but Wayback has it. Does this count as posted maintenance?

Wayback helps but Twitter/X is not "status page" per their terms. Read carefully: "posted to our status page at status.contabo.com." Social media doesn't count. The timing is suspicious though -- 8 minutes after your incident, they knew enough to tweet, but never updated the formal page? That suggests they knew and chose the informal channel specifically to create wiggle room.

I'd include the Wayback link anyway. It shows knowledge of incident they later denied. Doesn't win the SLA claim directly but paints a picture if you escalate to arbitration or chargeback threat.

Response time SLA: most providers bury this so deep nobody knows it exists. Contabo's is 24 hours for critical, right? 65 days is absurd. But good luck finding a lawyer in Dublin who cares about €4 VPS credit.

I am in São Paulo, use Contabo there. No SLA claim, no monitoring like you all. But I read this and check my own uptime. 99.3% in July. Many disconnect, many slow. I think is Brasil telecom problem, not Contabo. But how I know? I have no MTR, no logs, nothing. This thread make me feel I should start logging before I migrate to selfhost like I plan.

Vultr in São Paulo is better? Same price almost. Hetzner no Brasil, latency to Ashburn too high for my use.