How to request hardware diagnostics

Quick answer: If you suspect failing hardware (disk errors, ECC memory events, random reboots, thermal issues), open a support ticket with the symptoms and log excerpts. Cloud2Y engineers run diagnostics — some checks need the server taken offline (e.g. full memory tests), so a maintenance window may be agreed. Failed components are replaced.

Overview

Hardware rarely fails without warning: SMART attributes degrade, ECC counters tick up, kernel logs fill with I/O errors. Catching these early converts an outage into a planned disk swap. This article covers what to collect and how the diagnostics process works.

Before you start

• Collect evidence from the OS:

  dmesg -T | grep -iE 'error|fail|mce|i/o'
  journalctl -p err -S -48h
  smartctl -a /dev/sda        # per disk; via storcli/MegaCli behind HW RAID
  cat /proc/mdstat            # software RAID state

• Note patterns: when reboots/freezes happen, what changed recently.
• Plan for a possible maintenance window if offline tests are needed.

Step-by-step guide

1. Open a ticket: server ID/IP, symptoms, when they started, and the log excerpts above.
2. Engineers review remotely first — IPMI sensors, event logs, controller state.
3. If needed, agree a window for offline tests (memtest for RAM, extended disk tests).
4. Get the verdict in the ticket: component healthy, degraded or scheduled for replacement.
5. For disk swaps in a redundant RAID, replacement is usually done live; the array rebuilds afterwards — watch its state.

Common issues

• Reporting "server is slow" without data — slowness is usually software; logs and SMART output separate the two quickly.
• Ignoring a degraded array — with one mirror half gone you are one failure from data loss; report it immediately.
• Skipping backups before offline tests — diagnostics are non-destructive, but any maintenance is a good backup trigger.

When to contact support

Immediately on: repeated unexplained reboots, growing reallocated/pending sector counts, ECC error events, RAID degradation, or temperature alarms. Hardware health is Cloud2Y's side of the support scope — you will not be charged for failed-part replacement.

Frequently asked questions

Is hardware replacement paid?

No — diagnosing and replacing failed components is part of the Cloud2Y dedicated service. You only handle the OS side, such as watching a RAID rebuild afterwards.

Will diagnostics take my server offline?

Remote checks such as IPMI sensors and log reviews run live. Deep tests like full memory testing require the server offline, so a maintenance window is agreed with you first.

What evidence should I attach?

Attach dmesg or journalctl error excerpts, SMART output for each disk, the RAID state and a short description of when the symptoms occur and what changed recently.

Related articles

• How to monitor dedicated server performance
• How to request remote hands support
• How to set up RAID on a dedicated server
• Dedicated server SLA and support scope

Need a hand? Contact Cloud2Y support →