Setting the file. One moment.
Chapter 15 · Azure Diagnostics
Subchapter 15.27
troubleshooting/compute/references/cannot-connect-to-vm.mdMarkdown3 KBView on GitHub
Router for Azure VM connectivity issues. Determine OS first: Windows usually means RDP/3389 and TermService; Linux means SSH/22 and sshd; other images usually use SSH but may need current docs.
| Signal | Category | Reference |
|---|
| can’t RDP, timeout, black screen, internal error | RDP | rdp-connectivity.md |
| can’t SSH, refused, permission denied, publickey | SSH | ssh-connectivity.md |
| NSG, public IP, NIC, routes, effective rules | Network | network-connectivity.md |
| Windows Firewall, iptables, firewalld, UFW | Guest firewall | firewall-blocking.md |
| VM agent, Run Command timeout, Serial Console, boot diagnostics | VM agent/tools | vm-agent-not-responding.md |
| password, credentials, access denied, CredSSP, account expired | Credential/auth | credential-auth-errors.md |
| TermService, RDP disabled, changed port, TLS cert, NLA, licensing | RDP service/config | rdp-service-config.md |
Run before az vm user update, az vm user reset-ssh, az vm user reset-remote-desktop, az vm run-command invoke, or any extension-backed operation.
az vm get-instance-view --name <vm> -g <rg> \
--query "instanceView.{power:[statuses[?starts_with(code,'PowerState/')]][0][0].code,prov:[statuses[?starts_with(code,'ProvisioningState/')]][0][0].code,agent:vmAgent.statuses[0].displayStatus}" -o json
az vm extension list --vm-name <vm> -g <rg> \
--query "[].{name:name,state:provisioningState}" -o table| Check | Safe | Unsafe |
|---|---|---|
| Power | PowerState/running | other, missing, or query error |
| Provisioning | ProvisioningState/succeeded | creating/updating/deleting/failed, missing, or query error |
| VM agent | Ready | not ready, null, missing, or query error |
| Extensions | all Succeeded or none | creating/updating/deleting/failed |
If unsafe: stop, name the failed check, and use non-agent options such as Serial Console, offline repair VM, or Portal actions. If transient, wait and rerun checks only.
Check Resource Health, then offer restart or redeploy only with user approval. For broad docs, use Windows RDP (opens in a new tab), Linux SSH (opens in a new tab), Windows VM hub (opens in a new tab), or Linux VM hub (opens in a new tab).