VMware Horizon disconnects frequently: how can you distinguish client, network, Agent, and Connection Server problems?
Correlate Client, Agent, Connection Server, and firewall timestamps, then verify the display protocol, loss and jitter, MTU, proxy path, and session timeouts.

Establish a factual baseline with “client logs” and “Horizon Agent logs”, then move to “Connection Server logs”. Use logon, desktop establishment, protocol, resource and application experience as the evidence chain and avoid changing production before the fault boundary is clear.
Risk boundaries and prerequisites
Start by checking client logs, Horizon Agent logs and Connection Server logs independently. A healthy result for one of these checks does not establish the state of the entire service path.
Evidence and safe checks
| Check | Why it matters | Recommended action |
|---|---|---|
| 01 · client logs | Checking client logs helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of client logs. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
| 02 · Horizon Agent logs | Checking Horizon Agent logs helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of Horizon Agent logs. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
| 03 · Connection Server logs | Checking Connection Server logs helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of Connection Server logs. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
| 04 · Blast/PCoIP path | Checking Blast/PCoIP path helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of Blast/PCoIP path. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
| 05 · packet loss, jitter and MTU | Checking packet loss, jitter and MTU helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of packet loss, jitter and MTU. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
| 06 · firewall session timeout | Checking firewall session timeout helps establish whether the failure is isolated to a specific authentication, session or desktop stage; compare it with a known-good baseline and the incident timeline. | Record the actual state of firewall session timeout. Correlate client, authentication, agent and transport logs and reproduce on one controlled desktop. |
ping -t connection-server.corp.example
pathping connection-server.corp.exampleInvestigation and change sequence
- Capture a baseline for client logs and Horizon Agent logs and compare it with a known-good system.
- Correlate evidence from Connection Server logs and Blast/PCoIP path to distinguish the direct fault from dependent symptoms.
- Only if evidence points to packet loss, jitter and MTU, plan a controlled change with backups, a change window and rollback triggers.
- Finally check firewall session timeout and the real business workflow. Retest representative sign-in, application access and session or file redirection from the actual user location.
Validation and rollback evidence
- Retest representative sign-in, application access and session or file redirection from the actual user location.
- Recheck packet loss, jitter and MTU and firewall session timeout against the expected business outcome.
- Retain initial configuration, event evidence, validation results and rollback triggers for production changes.
Common actions that increase risk
- Assuming client logs is sufficient while overlooking Horizon Agent logs and dependent checks.
- Changing several settings before preserving the original evidence for Connection Server logs.
- Success on one desktop does not prove all pools and access locations are healthy.
Common questions and acceptance criteria
What should be checked first?
Begin with client logs and Horizon Agent logs. Capture reproducible errors and protect recoverable data before changing production.
What proves the issue is resolved?
Retest representative sign-in, application access and session or file redirection from the actual user location.
What evidence should be retained?
Retain original logs, configuration snapshots, the approved change scope, business verification and rollback conditions.
