Network problems require a layered diagnostic method
Network problems are notoriously intermittent and cross-cutting — a user who “can’t connect” could be hitting a VLAN misconfiguration, a firewall rule, a flapping VPN tunnel, or a DNS failure. The discipline that resolves network incidents fastest is the OSI layer approach: verify physical and data-link (cable, port, VLAN), then network (routing, firewall), then application (DNS, service). Jumping straight to “the firewall is blocking it” without confirming the lower layers wastes hours.
This article covers the network problems I see most in enterprise and MSP environments: VLAN and segmentation misconfiguration, firewall rule conflicts, VPN tunnel flapping, and DNS resolution failures. For the broader network management framework, see the companion guide on network infrastructure management.
Problem 1: Intermittent connectivity from VLAN misconfiguration
Intermittent connectivity between two hosts on the same switch often stems from a VLAN mismatch — a port assigned to the wrong VLAN, or a trunk missing the VLAN in its allowed list. The symptom is partial: some traffic works, some does not, and it changes as devices move.
The diagnostic sequence: confirm the port VLAN assignment on both ends with the switch CLI, verify trunk ports allow the VLAN (`show interfaces trunk`), and check that inter-VLAN routing points to the correct gateway. A VLAN present on an access port but missing from the trunk between switches produces exactly this intermittent pattern.
! Access port VLAN assignment
show interfaces GigabitEthernet1/0/24 switchport
! Trunk allowed VLANs
show interfaces trunk
! VLAN existence and interface membership
show vlan brief
! Inter-VLAN routing (SVI / gateway)
show ip interface brief | include VlanProblem 2: Firewall rule conflicts and shadow rules
A firewall that blocks traffic it should allow — or allows traffic it should block — usually has a rule conflict: a broad deny rule above a specific allow, or a legacy “shadow” rule that overrides the intended policy. As rule bases grow over years, these conflicts accumulate and become hard to reason about.
The diagnostic path is the firewall rule lookup (packet-tracer on Fortinet, packet flow on Sophos, the rule hit counters on any platform): trace the specific source, destination, and port to see which rule matched. If a deny rule matched before the intended allow, reorder or refine the rules. Audit the rule base periodically and remove rules with zero hits — a bloated rule base is both a security and a reliability risk.
| Symptom | Likely cause | Action |
|---|---|---|
| Traffic blocked unexpectedly | Broad deny above specific allow | Reorder rules; use packet tracer |
| Traffic allowed unexpectedly | Legacy shadow rule overriding policy | Remove or refine the shadow rule |
| Intermittent block | Failover policy differs across nodes | Sync rule base across HA pair |
| Rule works for one host only | Source object too narrow | Expand the source object or use groups |
Problem 3: VPN tunnel flapping
A site-to-site VPN tunnel that flaps (connects, drops, reconnects) is usually caused by one of: a Phase 1 / Phase 2 lifetime mismatch between peers, an MTU or fragmentation problem, or an unstable underlying internet link. The symptom is intermittent reachability across the tunnel with no clear pattern.
The diagnostic path: check the tunnel logs on both peers for Phase 1 and Phase 2 negotiation — a lifetime mismatch shows as repeated rekey failures. Verify the IKE and IPsec proposals match on both ends. For MTU, test with ping using the DF flag (`ping -f -l 1472`) to find the path MTU and adjust the tunnel MTU or enable TCP MSS clamping.
- Match Phase 1 and Phase 2 lifetimes and proposals on both peers
- Check tunnel logs for repeated rekey failures
- Test path MTU with ping -f -l 1472 and enable MSS clamping
- Verify the underlying internet link is stable, not just the tunnel
Problem 4: DNS resolution failures
DNS failures masquerade as network problems — a user who “can’t reach the server” often cannot resolve the server’s name. The symptom is reachability by IP but not by hostname, or intermittent resolution as the client falls back to a different DNS server.
The diagnostic sequence: confirm the client’s DNS server configuration (`ipconfig /all`), test resolution against the configured server (`nslookup server.domain 10.0.0.53`), and check the DNS server’s forwarders and zone health. A common cause is a stale or misconfigured forwarder, or a DNS scavenging setting that removed a valid record. AD-integrated DNS with overly aggressive scavenging is a frequent culprit.
# Client DNS configuration
Get-DnsClientServerAddress | Where-Object ServerAddresses | Format-Table
# Test resolution against a specific DNS server
Resolve-DnsName -Name "app.contoso.com" -Server 10.0.0.53 -Type A
# Check forwarders and zone on the DNS server
Get-DnsServerForwarder
Get-DnsServerZone | Where-Object IsReverseLookupZone -eq $false |
Select-Object ZoneName, ZoneType, IsIntegrated, AgingEnabledTrade-offs and next steps
Network design balances segmentation against complexity. More VLANs improve security and broadcast isolation but add routing and firewall complexity; a flat network is simple but offers no containment. The right balance is segmentation aligned to trust zones (user, server, guest, management) with documented routing and a rule base that is audited, not accumulated.
When problems recur, the next step is usually to mature the monitoring and documentation: deploy PRTG or equivalent for proactive alerting before users notice, maintain an accurate network diagram and IP plan, and audit the firewall rule base on a schedule. For the broader management framework, see the network infrastructure management guide.
- Segment by trust zone (user, server, guest, management) with documented routing
- Audit the firewall rule base on a schedule; remove zero-hit rules
- Deploy proactive monitoring (PRTG) so you know before users do
- Maintain an accurate network diagram and IP plan
Lessons learned
- Diagnose bottom-up: physical and data-link before network and application
- Use packet tracer / rule hit counters to find the matching firewall rule
- Match VPN Phase 1/Phase 2 lifetimes and test path MTU for flapping tunnels
- Confirm DNS before assuming a connectivity problem
Written by
Jamshed Alam
Global MSP & IT Infrastructure Consultant with 20 years across Exchange, Microsoft 365, Azure, security and automation. Available for project and contract work worldwide.