VMFS Recovery™
Recover data from damaged or formatted VMFS disks or VMDK files
Last updated:

Restarting Management Agents ESXi: Every Method, Every Warning, and When Restart ESXi Management Agents Is the Right Move

In VMware ESXi, the management agents (hostd and vpxa) are background services that let administrators interact with the host through the vSphere Client, vCenter Server, and APIs. When these agents stop responding, hosts may appear as “disconnected” in vCenter, vMotion operations may fail, or VMs may not be manageable through the GUI. Restarting the affected service can restore management access, but first check for network, storage, or resource problems that may be causing the symptoms.

This guide covers the main restart methods: GUI options in the Host Client, CLI commands through ESXi Shell and SSH, and DCUI (Direct Console User Interface) workflows. It also explains when to use each method, how to avoid disrupting running workloads, and what precautions to take before restarting. Choose the method that fits your host configuration and verify connectivity afterward.

What Are ESXi Management Agents?

hostd: The Core Host Management Daemon

  • ⚙️ hostd is the primary management daemon running on every ESXi host.
  • 🖥️ Handles local host operations: VM power on/off, hardware configuration changes, storage management, snapshots, and VMware Host Client communication.
  • ❌ If hostd hangs or crashes, the host becomes unresponsive to vSphere Client, vCenter, and web browser connections.
  • Important: A management-agent failure does not necessarily mean guest VMs have stopped. Check their actual state and connectivity before taking further action.

vpxa: The vCenter Agent

  • 🔗 vpxa (vCenter Proxy Agent) bridges communication between ESXi hosts and vCenter Server.
  • 📡 Relays commands from the vpxd service in vCenter down to hostd.
  • ❌ If vpxa fails, the host shows as “Not Responding” or “Disconnected” in vCenter.
  • ✅ Direct connections via VMware Host Client may still work. Restarting vpxa may restore vCenter connectivity. Before restarting it, check whether a network or host failure is responsible.

Other Management Services

  • 🛡️ fdm → vSphere HA agent, manages VM restart after host failure in HA clusters.
  • 🔍 sfcbd → CIM/WBEM broker for hardware monitoring; CIM is deprecated in ESXi 8.0.
  • 🌐 slpd → Service Location Protocol daemon for discovery; SLP is deprecated in ESXi 8.0.
  • 📊 SNMP → Network monitoring agent.
  • ⚠️ Best practice: Restart these individually using the init.d mechanism — never use services.sh restart in production, as it disrupts all services at once.
Tip: ESX vs ESXi

📊 ESXi Management Agent Reference Table

Treat the impact column as a description of intended management-plane behavior, not a guarantee of uninterrupted VM operation. Broadcom KB 320280 warns that restarting management agents can disrupt running tasks, affect guest performance, or cause instability that requires a host reboot. Check vSAN, LACP, NSX, and shared-graphics restrictions before selecting a restart method.

ServiceBinary / CommandRoleRestart Impact on VMs
hostd/etc/init.d/hostd restartLocal host management, VM operationsManagement access is interrupted; active tasks and guest performance may be affected
vpxa/etc/init.d/vpxa restartvCenter Server communicationvCenter connectivity is interrupted; verify reconnection and check active tasks
fdm/etc/init.d/vmware-fdm restartvSphere HA failover managementHA agent is temporarily unavailable; verify HA health after the restart
sfcbd/etc/init.d/sfcbd-watchdog restartCIM hardware monitoringCIM hardware monitoring is temporarily unavailable
slpd/etc/init.d/slpd restartSLP service discoverySLP discovery is temporarily unavailable

When to Restart Management Agents on ESXi

Symptoms That Indicate a Management Agent Restart Is Needed

  • 🖥️ ESXi host shows “Not Responding” or “Disconnected” in vCenter while VMs continue running → may indicate a vpxa failure or lost connectivity between host and vCenter.
  • vSphere Client cannot connect directly to the host (timeout / “unable to connect” error) → may indicate a hostd failure or a connectivity issue.
  • VM actions (power on/off, snapshot, vMotion) fail or hang in vCenter → investigate hostd responsiveness along with storage, network, and resource conditions.
  • 📌 Error: “Virtual machine creation may fail because the agent is unable to retrieve VM creation options from the host” → a hostd restart may help if an unresponsive agent is the cause.
  • 👀 VMs appear inaccessible in vCenter but are visible/running when connected directly to the host → hostd/vpxa desynchronization.
  • vCenter reports the host entering or exiting maintenance mode without administrator action → review event history and automation before attributing it to vpxa.
Note: ESXi Repair

When NOT to Restart Management Agents

⚠️ Broadcom KB 320280 explicitly warns against using services.sh restart or the DCUI Restart Management Agents option in certain environments. Doing so can cause immediate outages:

  • 📦 vSAN environments → Restarting all services disrupts the vSAN storage layer. Restart hostd and vpxa individually only.
  • 🔗 LACP (Link Aggregation Control Protocol) → services.sh restart resets networking stack, breaking LACP uplinks. Restart individual services only.
  • 🛡️ NSX configured hosts → Restarting all services also restarts NSX VIBs, disrupting firewall and overlay networking. Check with:
esxcli software vib list | grep -E 'nsx|vsipfwlib'

If NSX is present, follow the documented procedure for the affected service and restart services individually. Do not use a broad management-agent restart.

  • 🎨 Shared graphics (VMware View with vGPU, vDGA, vSGA) → services.sh restart stops xorg, which crashes all VDI sessions. Restart only the affected service individually, following the procedure for that host configuration.
Note: ESXi Format Disk

Method 1: Restart ESXi Management Agents via DCUI

When to Use DCUI

  • 🖥️ The Direct Console User Interface (DCUI) can provide console access when network connectivity is unstable or SSH is disabled.
  • 🔌 Requires physical console access (monitor + keyboard) or remote console via iDRAC, iLO, or IPMI.
  • ⚙️ The DCUI Restart Management Agents option restarts hostd and vpxa sequentially.

Step‑by‑Step: DCUI Restart

  1. 1. 💻 Connect to the ESXi host console — either physically or via remote console (iDRAC/iLO/IPMI).
  2. 2. ⌨️ Press F2 to enter Customize System / View Logs. Enter root credentials when prompted.
  3. 3. 🔍 Use arrow keys to navigate to Troubleshooting Options → press Enter.
  4. 4. 🔧 Select Restart Management Agents → press Enter.
  5. 5. ⚡ Press F11 to confirm. Console shows:
  • Stopping management agents. Done.
  • Starting management agents. Done.
    1. 6. ✅ Press Enter after completion, then Esc to return to the DCUI main screen.
    2. 7. 🔄 Verify the host reconnects to vCenter within 1–2 minutes. Check the vCenter events log for reconnection.

    ⚠️ Warning: Do not use this method if the host is part of:

    • 📦 vSAN clusters → disrupts storage layer.
    • 🔗 LACP uplinks → breaks teamed NICs.
    • 🛡️ NSX deployments → restarts NSX VIBs, disrupting firewall/overlay networking.
    • 🎨 Shared graphics (VDI with vGPU/vDGA/vSGA) → stops xorg, crashing VDI sessions.

    👉 In these environments, use individual service restarts (Method 2) instead.

    Method 2: Restart ESXi Management Agents via SSH

    To restart ESXi management agents over SSH, restart the affected service individually, then check its state and host connectivity.

    Enable SSH on the ESXi Host

    SSH is disabled by default. Enable it using one of these methods:

    • 🖥️ DCUI → Troubleshooting Options → Enable SSH → Press Enter.
    • 🌐 vSphere Client → Host → Actions → Services → Enable Secure Shell (SSH).
    • ⚙️ VMware Host Client → Host → Manage → Services → TSM-SSH → Start.

    Restart individual management agents through SSH (check host configuration first)

    1. 1. 🔑 Connect to the ESXi host with an SSH client (PuTTY, OpenSSH, Windows Terminal) using root credentials.
    2. 2. Run the following commands depending on the issue:
  • Restart hostd (fixes local management / vSphere Client issues):
  • /etc/init.d/hostd restart
  • Restart vpxa (fixes vCenter “Not Responding” / “Disconnected”):
  • /etc/init.d/vpxa restart
  • If both agents need to be restarted, run each command separately in the order documented by Broadcom:
  • /etc/init.d/hostd restart
    /etc/init.d/vpxa restart
  • For an HA-agent issue that requires an FDM restart, Broadcom KB 409095 documents the following service script for ESXi 7.0 U3 and later:
  • /etc/init.d/vmware-fdm restart
    1. 3. Verify that services are running. Broadcom KB 405154 documents these status commands for ESXi 7.x and 8.x:
    /etc/init.d/hostd status
    /etc/init.d/vpxa status
    1. 4. Check management connectivity and review the hostd log if a service does not recover. Broadcom KB 320280 identifies this log; the following command shows its latest 100 lines:
    tail -100 /var/log/hostd.log

    services.sh restart: When It Is Safe and When It Is Not

    • ⚡ /sbin/services.sh restart restarts all management services at once. Faster, but riskier.

    Check the following before considering a broad restart: The absence of these features does not guarantee a disruption-free restart.

    • Standalone ESXi hosts.
    • No vSAN, no LACP, no NSX, no VMware View shared graphics.

    Never use:

    • 📦 vSAN → disrupts storage.
    • 🔗 LACP → breaks uplinks.
    • 🛡️ NSX → restarts firewall/overlay VIBs.
    • 🎨 Shared graphics (VDI) → stops xorg, crashes sessions.

    👉 If unsure whether NSX is installed, check first:

    esxcli software vib list | grep -E 'nsx|vsipfwlib'

    If NSX VIBs are present → restart only the affected service individually. Do not use a broad restart on the restricted configurations above. If individual restarts fail, investigate the logs and consult the procedure for your exact ESXi version before planning further maintenance.

    Method 3: Restart Management Agents ESXi via VMware Host Client

    Step‑by‑Step: ESXi Host Client Service Restart

    The VMware Host Client provides a GUI‑based way to restart individual services — ideal when SSH is disabled and DCUI access is unavailable.

    1. 1. 🌐 Open a web browser → navigate to:
    https://[ESXi-host-IP-address]

    Enter root credentials to log in.

    1. 2. 📂 In the left navigation pane, click Host → Manage.
    2. 3. ⚙️ Select the Services tab. A full list of ESXi services appears with their current status.
    3. 4. 🔗 Locate vpxa (VMware vCenter Agent) → click the service name → click Restart in the Actions toolbar.
    4. 5. 🔧 Repeat for hostd (VMware ESXi Daemon / Host Agent) if direct connection issues persist.
    5. 6. ✅ Monitor the service status column — it should transition from Stopped → Running within seconds.

    📌 Note: Service names differ between Host Client and SSH:

    • In SSH → vpxa and hostd.
    • In Host Client → VMware vCenter Agent and VMware ESXi Daemon.

    Method 4: Restart ESXi Management Agents via PowerCLI

    Restart vpxa via PowerCLI for Multi‑Host Environments

    PowerCLI enables scripted agent restarts across multiple hosts — essential in environments with 10+ ESXi hosts, where per‑host DCUI or SSH access is impractical.

    Step 1. Connect to vCenter or ESXi Host

    Connect-VIServer -Server 192.168.1.10

    Step 2. List Services on Target Host (refresh data first)

    Get-VMHostService -VMHost 192.168.1.100 -Refresh

    Step 3. Restart vpxa on a Single Host

    $vpxa = Get-VMHostService -VMHost 192.168.1.100 -Refresh | Where-Object { $_.Key -eq "vpxa" }
    Restart-VMHostService -HostService $vpxa -Confirm:$false

    Step 4. Restart hostd on a Single Host

    $hostd = Get-VMHostService -VMHost 192.168.1.100 -Refresh | Where-Object { $_.Key -eq "hostd" }
    Restart-VMHostService -HostService $hostd -Confirm:$false

    Step 5. Restart vpxa Across All Disconnected Hosts in a Cluster

    Get-Cluster "ClusterName" | Get-VMHost | Where-Object { $_.ConnectionState -eq "NotResponding" } | ForEach-Object {
        $vpxa = Get-VMHostService -VMHost $_ -Refresh | Where-Object { $_.Key -eq "vpxa" }
        Restart-VMHostService -HostService $vpxa -Confirm:$false
        Write-Host "Restarted vpxa on: $($_.Name)"
    }

    ⚠️ Best Practice Notes:

    • 🛡️ Use PowerCLI for bulk operations only when you’re confident in service dependencies.
    • ✅ Always refresh service data before restart.
    • 🔍 Monitor vCenter events after execution to confirm host reconnections.

    After the Restart: Verification and Log Review

    Verify Host Reconnects to vCenter

    • ⏱️ After restarting management agents, the host should reconnect within 2–3 minutes.
    • 📊 In vCenter → Monitor → Events → look for:
      • “ESXi host connected”
      • “vSphere HA agent on [host] completed reconfiguration”
    • 💻 Via PowerCLI:
    Get-VMHost 192.168.1.100 | Select-Object Name, ConnectionState, PowerState

    Expected output:

  • ConnectionState: Connected
  • PowerState: PoweredOn
  • ⚠️ If the host does not reconnect within 5 minutes, check network connectivity:
  • ping [vCenter-IP]

    (Run from ESXi SSH shell).

    Review Management Agent Logs After Restart

    If the restart fixes the immediate issue but it recurs later, investigate logs for root cause:

    • 📖 hostd log (local management activity):
    tail -f /var/log/hostd.log
    • 🔗 vpxa log (vCenter communication):
    tail -f /var/log/vpxa.log
    • 🔍 Search for errors before restart:
    grep -i "error\|fatal\|warn" /var/log/hostd.log | tail -100

    Common root causes found in logs:

    • 📦 Storage I/O latency → hostd queues fill.
    • 🧠 Memory exhaustion → ESXi kernel OOM killer terminates hostd/vpxa.
    • 🌐 NFC (Network File Copy) overload → backup operations exhaust hostd connection pool.

    Escalation Path When Management Agent Restart Does Not Resolve the Issue

    If agents crash repeatedly:

    • 🧠 Check host memory → If over‑committed, increase RAM or reduce VM consolidation ratio.
    • 📦 Check storage latency → VMFS datastore latency >20 ms overloads hostd queues. Investigate HBAs, drivers, datastore health.
    • 🛠️ Check ESXi patch level → Some builds have known hostd memory leaks/crash bugs. Review Broadcom release notes.
    • 🔄 Final step → If unresolved, schedule a host maintenance window and reboot the ESXi host. A full reboot reinitializes kernel state and clears conditions that agent restarts cannot fix.

    Management Agent Restarts and VMFS Data Integrity

    Do Management Agent Restarts Affect Running VMs?

    • Restarting hostd or vpxa does not itself reboot the host. Broadcom still warns that agent restarts may disrupt active tasks, affect guest VM performance, or cause instability that requires a reboot.
    • ⚙️ The VMkernel continues scheduling VM execution independently of management agents.
    • Check guest network connectivity and storage I/O after the restart; uninterrupted operation is not guaranteed.
    • ⏱️ Temporary impact → management connectivity (vSphere Client, vCenter visibility, remote console) drops briefly and restores once agents restart.
    • ⚠️ Operational caveat → active snapshot, provisioning, migration, or backup tasks may be disrupted. Review their actual status and investigate failures before retrying.

    When Management Agent Issues Precede or Accompany VMFS Failures

    • 📦 Management agent crashes often correlate with storage I/O degradation.
    • 🛑 If hostd is overwhelmed by latency, snapshot consolidation may be interrupted → leaving orphaned delta VMDK files.
    • ⚠️ Severe storage events can interrupt VMFS metadata writes, leaving the datastore partially consistent.
    • 🔄 A forced host reboot after persistent hostd crashes may interrupt in‑flight VMFS operations → risking metadata corruption.

    Recovering VMFS and VMDK Data After Management Agent‑Related Storage Failures with DiskInternals VMFS Recovery™

    When VMFS datastores become inaccessible due to agent crashes or storage events, standard ESXi tools cannot repair VMFS‑level damage.

    🛡️ DiskInternals VMFS Recovery™ provides:

    • 💾 VMDK mounting without a running ESXi host.
    • 🔧 VMFS volume reconstruction from damaged or partially overwritten metadata.
    • 📂 VMX configuration file recovery from datastores that went offline during hostd crashes.
    • 🌐 Remote ESXi server connections via IP + credentials for direct datastore scanning once management connectivity is restored.

    Workflow:

    1. 1. Connect to the affected VMFS volume.
    2. 2. Run a full scan.
    3. 3. Locate VMX and VMDK files in the recovery browser.
    4. 4. 👀 Preview file integrity in read‑only mode.
    5. 5. 📤 Extract recovered files to a safe destination (local drive, network share, or FTP).

    Ready to get your data back?

    To start recovering your data, documents, databases, images, videos, and other files, press the FREE DOWNLOAD button below to get the latest version of DiskInternals VMFS Recovery® and begin the step-by-step recovery process. You can preview all recovered files absolutely for FREE. To check the current prices, please press the Get Prices button. If you need any assistance, please feel free to contact Technical Support. The team is here to help you Recover Deleted Datastore ESX!

    FAQ

    Related articles

    FREE DOWNLOADVer 4.28, WinBUY NOWFrom $699