vSAN iSCSI Target Latency and Application Timeouts Due to Workload Saturation on a Single Target
Article ID: 448560
Updated On:
Products
Issue/Introduction
High storage latency is observed along the iSCSI network path, causing performance degradation, timeouts, and operational impact for connected applications.
The guest operating system records connection timeouts, disk retries, or storage target resets for the affected iSCSI devices.
Performance degradation typically occurs during peak operational hours or periods of elevated application activity.
Standard infrastructure recovery steps—such as restarting application services, rebooting the active storage owner host, or migrating storage ownership to another node in the cluster—fail to provide sustained remediation.Network and physical storage fabric diagnostics indicate healthy component status, with no signs of physical hardware degradation, packet loss, or link errors.
- At the time of the latency is observed, below messages are observed in vmkwarning, where the tasks are getting queued up.
2026-06-10T23:27:32.809Z Wa(180) vmkwarning: cpu22:2099982)WARNING: vit: VitTaskQueueEnqueue:2080: Throttled: Skip enqueuing task for BE LUN vit-lun-193302322.3001 tasks for the LUN are in queue. Total 3012 tasks in queue, 4 LUNs have tasks in queue.
2026-06-10T23:28:35.055Z Wa(180) vmkwarning: cpu25:2099982)WARNING: vit: VitTaskQueueEnqueue:2080: Throttled: Skip enqueuing task for BE LUN vit-lun-193302322.2999 tasks for the LUN are in queue. Total 3010 tasks in queue, 2 LUNs
have tasks in queue - Based on the performance graph, a substantial increase in input/output operations per second (IOPS) is observable, during which throughput reaches maximum capacity while storage latency and queue depth increase correspondingly.
Environment
VMware vSAN (All versions)
Cause
The issue is caused by too many high-demand storage disks (LUNs) being grouped together under a single vSAN iSCSI target. All storage traffic from these disks must pass through one active network queue on a single ESXi host. When applications generate high numbers of input/output operations per second (IOPS), the host cannot process the incoming storage commands fast enough. This causes thousands of commands to back up in the driver queue, triggering built-in throttling, latency spikes, and connection timeouts.
Resolution
A vSAN iSCSI target can only be owned by one ESXi host at a time. Creating multiple targets allows the cluster to distribute the LUNs across different physical hosts. This balances the aggregate IOPS load, stops the storage queues from filling up, and eliminates the latency.
Feedback
