Packet loss investigation
Packet loss needs tests from both sides and a clear time window.
Steps
- Run ping and traceroute from multiple locations.
- Check interface errors and bandwidth use.
- Identify whether loss is inside server, rack, facility, or upstream.
- Escalate with evidence.
Useful information to include
- Source and destination
- MTR output
- Interface counters
Escalation
Contact Metal on Cloud support when the issue affects production, requires privileged access, involves billing, or needs a formal incident record.