Troubleshoot And Optimize practice questions

Generate a support bundle from the vSphere Client or examine the configured diagnostic partition / scratch partition using 'esxcli system coredump file get'.

The PSOD writes its core dump to the configured coredump partition, not the support bundle. Running `esxcli system coredump file get` on the host reveals the active dump file path and partition, letting the administrator retrieve the exact file VMware Support requires for crash analysis.

Compare Kernel Latency (KAVG) and Device Latency (DAVG/GAVG) to see if the delay is introduced inside the ESXi storage stack or by the storage array.

KAVG measures time inside the ESXi storage stack, while DAVG and GAVG reflect device and guest-observed latency. Comparing them isolates whether delay originates in the hypervisor stack or the array, pinpointing the source of the spikes.

Check the 'Max Limited' or 'CPU Limit' metric in the vSphere performance charts for the affected virtual machine.

The 'Max Limited' or 'CPU Limit' counter records the percentage of time the VM was ready but unable to run because a configured CPU limit was reached, directly identifying constraint imposed by DRS or administrator settings rather than contention.

Storage remains available as long as the primary site and the witness maintain data path connectivity, but vSAN health alarms will trigger due to management loss.

Management network loss does not automatically tear down the vSAN data path (VMkernel traffic), though monitoring and alarm propagation are impaired.

The destination datastore must have sufficient free capacity to accommodate the virtual machine's thick-provisioned or thin-provisioned footprint.

Space validation is mandatory; insufficient free space causes the migration task to fail immediately.

Both the source and destination hosts must have active storage accessibility (paths) to both datastores during the migration phase.

During Storage vMotion, the host reads from the source and writes to the destination simultaneously, requiring paths to both datastores.

The VM must not have snapshot files larger than 2TB if using older VMFS versions, and disk formats must be supported.

Snapshot chains and disk configurations must be supported by the migration engine (e.g., physical RDMs cannot be migrated without converting to VMDK).

Log into the VAMI (port 5480) or appliance shell via SSH to identify which partition is full and inspect log directories under /var/log/.

Logging into the VAMI or appliance shell via SSH directly satisfies the need to identify which partition is full before deleting anything. Inspecting /var/log/ reveals the specific log directories consuming root partition space, enabling safe, targeted cleanup rather than blind deletion that could break VCSA services.

MCTLSZ (Ballooned memory size)

MCTLSZ reports the amount of guest memory reclaimed by the balloon driver, so a sustained non-zero value directly evidences active reclamation under memory pressure, satisfying the stem's requirement for a primary pressure indicator in esxtop.

SWCUR (Current swap usage)

SWCUR shows the current swapfile consumption per VM, revealing that the host has already exhausted physical RAM and is actively paging VM memory to disk. Rising SWCUR is a primary indicator of severe memory pressure and active reclamation, directly satisfying the stem's requirement.

esxcli storage core adapter rescan --all

This command rescans all storage adapters on the ESXi host, refreshing paths to Fibre Channel, FCoE, and iSCSI targets.

The vCenter Statistics service (integrated into vpxd and the associated database backend).

The vCenter Statistics service, integrated into vpxd, periodically polls ESXi hosts for performance metrics and writes them to the vCenter database, which the vSphere Client reads to render graphs. Flatlines or missing historical data therefore indicate this collection or storage pipeline has failed, directly matching the stem's symptom.

pktcap-uw --uplink vmnicX --dir 0 -o /tmp/capture.pcap

pktcap-uw with --uplink targets a specific vmnic, and --dir 0 captures ingress traffic, letting the administrator inspect whether 802.1Q VLAN tags arrive on the trunk uplink. The -o flag writes the capture to a pcap file for analysis.

Set the ESXi host Power Management Policy to 'High Performance' (Static High).

Static High power policy pins CPU cores at maximum frequency, preventing the latency and jitter caused by frequency scaling and deep C-state wake transitions. This satisfies the stem's requirement for lowest possible latency and jitter on latency-sensitive workloads.

Reserve 100% of the virtual machine's guest memory to prevent hypervisor ballooning and swapping.

Reserving 100% of guest memory prevents the hypervisor from ballooning or swapping pages, both of which introduce unpredictable stalls. Eliminating those reclamation paths directly satisfies the stem's requirement to minimise latency and jitter for latency-sensitive applications.

Configure the virtual machine's Latency Sensitivity setting to 'High' in vSphere Client.

Setting Latency Sensitivity to 'High' enables CPU affinity and disables hypervisor scheduling tricks that add jitter, reserving dedicated physical cores for the VM. This directly satisfies the latency-sensitive, low-jitter requirement by eliminating CPU scheduling delays.

If the disk has failed permanently, replace the physical drive and rebuild or recreate the vSAN disk group, allowing vSAN to automatically rebuild missing components based on storage policies.

Replacing a permanently failed drive and recreating the disk group restores the capacity and fault domain the cluster lost, letting vSAN's automatic rebuild regenerate absent components to satisfy the storage policy's failures-to-tolerate. This directly addresses the degraded state by removing the failed device and re-establishing the disk group as a rebuild target.

Check if the host containing the disk group is in maintenance mode or has been disconnected from the cluster.

A host in maintenance mode or disconnected from the cluster makes its disk group components inaccessible, producing exactly the Degraded state and absent components reported. Verifying host connectivity and maintenance status therefore identifies whether the alarm is transient rather than genuine disk failure.

Review the vSAN Health service in vSphere Client to identify specific unhealthy components, failed disks, or network partitions.

Reviewing the vSAN Health service in vSphere Client pinpoints the exact failing element behind the Degraded disk group — whether a physically failed disk, an absent component, or a network partition isolating hosts. This satisfies the investigation constraint by distinguishing hardware failure from connectivity loss before any remediation, since each cause demands a different corrective action.

esxcli storage nmp path list

This command lists every path to each LUN, showing its operational state (active, dead, standby), the assigned PSP, and the array controller details, letting the administrator pinpoint why the FC LUN is flapping between active and dead paths.

TPS (Transparent Page Sharing), followed by Ballooning (vmmemctl), Memory Compression, and finally Swapping.

ESXi reclaims memory in escalating order: transparent page sharing first, then ballooning via the vmmemctl driver, then memory compression, and finally swapping to disk. This sequence satisfies the stated order, exhausting cheaper techniques before the severe performance cost of swapping.

vSphere Client Performance Charts (Datastore and VM storage views)

vSphere Client Performance Charts expose datastore-level and per-VM storage metrics such as latency, IOPS and throughput over time, letting the administrator correlate sluggish responsiveness with a specific VMFS datastore or virtual disk without touching the ESXi shell.

esxtop (pressing 'u' then 's' to view storage statistics and latency metrics like DAVG, KAVG, and QAVG).

esxtop's storage views ('u' for device, 's' for datastore) report per-device DAVG, KAVG and QAVG latency, isolating whether sluggishness originates in the guest, the VMkernel queue or the physical path — the granularity Performance Charts alone cannot provide.

esxcli storage nmp path list and esxcli storage core device list

esxcli storage nmp path list reveals each device's Pluggable Storage Architecture path state (active, standby, dead), while storage core device list shows device and APD/PDL status — pinpointing a failed or thrashing path behind the VMFS sluggishness.

%RUN (Percentage of time scheduled on a physical CPU)

%RUN indicates how much CPU time the world was actively running on physical cores.

%USED (Percentage of physical CPU used)

%USED shows the actual physical CPU resources consumed by the world (VM).

esxcli network vswitch dvs lacp status get or inspecting lacp logs in /var/log/

The `esxcli network vswitch dvs lacp status get` command directly reports LACP negotiation state per uplink, revealing whether the port channel is down due to mismatched LACP mode, timeout, or partner information. This satisfies the stem's requirement to inspect LACP state and negotiation status on the ESXi host.

Press 'd' while running esxtop to display the physical adapter / disk controller statistics view.

The 'd' interactive command switches esxtop to the disk adapter view, showing physical HBA and disk controller statistics such as queue depth and latency. This provides the storage performance data needed to diagnose the ESXi host's issue.

Press 'v' while running esxtop to display the virtual disk statistics view.

Pressing 'v' switches esxtop to the virtual disk view, which reports per-virtual-machine disk metrics such as latency, IOPS and queue depth. This directly satisfies the stem's requirement for storage performance statistics, letting the administrator identify which VM's virtual disks are causing the ESXi host's storage bottleneck.

A complete network partition occurred where the surviving hosts cannot communicate with the isolated host and datastore heartbeats were also unavailable.

Without network heartbeats or datastore heartbeats, HA cannot reliably verify host state or safely initiate restarts.

The virtual machines were running with manual restart priority set to 'Disabled' or excluded from HA protection in the cluster settings.

Individual VMs can have their HA restart priority set to disabled, preventing automated restarts.

Admission control is enabled, and the remaining healthy hosts in the cluster lack sufficient unreserved capacity to power on the failed VMs.

Admission control blocks restarts if failover resources are insufficient.

Select the vDS -> Monitor -> Uplink / Topology / Physical Switch view to inspect CDP/LLDP neighbor details.

CDP/LLDP discovery reveals physical switch port IDs, switch names, and native VLAN configurations directly within the vSphere Client.