joshfixes.it
← All guides

Networking / demonstration

Debugging intermittent network connectivity

A demonstration method for turning a vague drop-out into timestamped, layer-specific evidence.

Published
30/08/2026
Updated
30/08/2026
Reading
8 minutes

Demonstration guide — commands and outcomes are illustrative. Review before use.

Define “intermittent”

Record which client, destination, protocol and time window are affected. A useful first result distinguishes name resolution, local reachability, routing and application response.

# Replace documentation addresses with approved test targets
ping -c 5 192.0.2.1

Command: Run small, timestamped samples. An endless ping creates noise and can hide the shape of the failure.

Test one boundary at a time

  1. Confirm link state and local interface counters.
  2. Test the first routed hop.
  3. Resolve the destination independently of connecting to it.
  4. Test the application protocol.
  5. Capture traffic only across the smallest relevant window.

Interpret the evidence

Loss at every destination suggests a local link or first-hop problem. Successful IP tests with failed names point toward DNS. One failing application with healthy transport needs application-layer investigation.

Tip: Keep clocks synchronised before correlating client, switch, firewall and service logs.

Verify

Repeat the same controlled sample after the change and retain both results. “It seems stable now” is not a repeatable exit criterion.