--- title: "Support policy" canonical_url: "https://3nv.io/legal/support-policy" --- > Index of every page on this site: [llms.txt](https://3nv.io/llms.txt) ## In plain words - Hardware failures get a response at any hour, every day. Post in your shared Slack or Teams channel, or email [support@3nv.io](mailto:support@3nv.io) with "URGENT" in the subject. - Everything else is handled in business hours: Monday to Friday, 9:00–17:00 Tallinn time, except Estonian public holidays. - We support the hardware, firmware, network and facility. The operating system and everything above it is yours, unless we agree a separate scope. - You talk to the engineers who build and run your servers. - After a serious outage, you get a written report within 5 business days. - We give 10 business days' notice of planned maintenance. - There is no uptime figure before commissioning. Service levels for your deployment are agreed after it. Version 1, 26 October 2026. ## 1. Scope 1.1. This policy applies to Reserve and Own. For Colocate it covers the facility, power and our network only. It is part of the [Terms of Service](https://3nv.io/legal/terms-of-service). 1.2. **We support:** the facility, power, cooling, hardware, firmware, the management controller, our network and IP transit, and replacing failed parts. 1.3. **We don't support:** the operating system, drivers, CUDA, models, data, applications or your own network setup inside the server. We can scope and quote work above the operating system separately. ## 2. How to reach us 2.1. **Hardware failures, any hour:** post in your shared support channel (Slack Connect or a Microsoft Teams shared channel, set up at handover), or email [support@3nv.io](mailto:support@3nv.io) with "URGENT" in the subject. Both page the engineer on call. Use "URGENT" only for hardware failures. 2.2. **Everything else, business hours:** your shared support channel or [support@3nv.io](mailto:support@3nv.io). Business hours are Monday to Friday, 9:00–17:00 Tallinn time, excluding Estonian public holidays. 2.3. **Who can ask.** At handover you name the people who may raise urgent requests and approve work on your server. Only they can page us or approve a power-off. You can change the list at any time in writing. 2.4. We support you in English or Estonian. 2.5. Other addresses: [sales@3nv.io](mailto:sales@3nv.io) for new deployments and quotes, [security@3nv.io](mailto:security@3nv.io) for security issues, [abuse@3nv.io](mailto:abuse@3nv.io) for abuse reports, [privacy@3nv.io](mailto:privacy@3nv.io) for personal data requests and [legal@3nv.io](mailto:legal@3nv.io) for authorities. 2.6. Tell us the server, what you see and when it started. If it's a hardware failure, say so in the first line. ## 3. What counts as a hardware failure 3.1. A server that won't boot. A failed GPU, drive, power supply, fan or memory module. A lost link between GPUs, or a lost link to the cluster fabric. 3.2. Loss of power or cooling to your rack, or loss of our network to your server. 3.3. Anything else, including performance questions, firmware updates you request and operating system problems, is handled in business hours. ## 4. Severity and response | Severity | Meaning | First response | Updates | | --- | --- | --- | --- | | 1 | Server down, a hardware failure stopping your work, or a suspected security incident on our side | 1 hour, any time | Every 2 hours until fixed | | 2 | Hardware degraded, server still running | 4 hours, any time | At least once a day | | 3 | Question or request | Next business day | When there's news | - A failure that stops your job is Severity 1, even if the server is still on. Example: one GPU fails in an 8-GPU training run. - A suspected security breach of our hardware, management controller, network or facility is Severity 1. Also tell [security@3nv.io](mailto:security@3nv.io). - You choose the severity. If we think it's different, we tell you why. - "Next business day" means by 17:00 Tallinn time on the next business day. 4.1. "First response" means an engineer has looked at the issue and replied. It doesn't mean the issue is fixed. The clock starts when your message reaches your channel or [support@3nv.io](mailto:support@3nv.io). 4.2. These are targets, not commitments with credits. We don't publish fix times: they depend on the part. Service levels with remedies are agreed per deployment after commissioning. 4.3. **Escalation.** Not happy with progress? Write "escalate" in your channel. One of the founders replies within 1 hour for Severity 1, at any time, and the next business day otherwise. Their direct contacts are in your handover pack. 4.4. **Incident reports.** After a Severity 1 incident, we send a written report within 5 business days. It says what happened, why, what we changed, and whether we met our targets. ## 5. When hardware fails 5.1. We diagnose through the management controller and replace the failed part. We keep common spares (drives, fans, power supplies and memory) on site. GPUs and boards are replaced under the vendor warranty. Your quote lists the spares we hold for your deployment and a target time to replace each part. 5.2. Failed drives are destroyed on site and never returned to a vendor. On Own and Colocate the drive is yours: we destroy it on site unless your order form says to hand it back. 5.3. For Own, we handle warranty claims with the vendor for you. Once the warranty has ended, you pay for replacement parts at cost. Our time to swap them is covered by your management fee. 5.4. Every login to your server's management controller or console is logged, with who and when. 5.5. Some fixes need your server powered off. We ask one of your named contacts first, unless you've told us in advance that we may. While we wait for an answer, the response and update clocks are paused. 5.6. Outside business hours, the data centre's on-site staff may do the physical work on our ticket, following our instructions. Our engineer stays on the case until it's fixed. ## 6. Maintenance 6.1. We give at least 10 business days' notice of planned maintenance that affects your server. If the work needs a reboot or a power-off, we agree the time with you. We don't start it during a job you've told us is running, unless safety requires it. 6.2. Urgent security or safety work may need shorter notice. We tell you before we start if we can, and always within 1 hour of starting. 6.3. We post incidents and maintenance in your shared channel. ## 7. Service levels 7.1. No uptime figure applies before commissioning. 7.2. After commissioning, we agree service levels and any service credits for your deployment and record them in writing. Where they conflict with this policy, they win. 7.3. Deployment service levels say: - how we measure availability; - target replacement times per part; - what's excluded, such as planned maintenance with notice, your own changes, and events beyond our control; - any service credits, and their cap; - how to claim them. Credit levels and their cap are agreed per deployment. You claim credits within 30 days after the month in which a target was missed. ## Sitemap See the full [sitemap](https://3nv.io/sitemap.md) for all pages.