Skip to main content
Migration Notice
We're migrating documentation from the old portal into this one. Some things may look a little different or out of place in the meantime — we know, and we're working to get it right. If something's unclear or doesn't look right, let us know.

Agent Dashboard — Disk Read Latency

What this chart shows​

Disk Read Latency (ms) — the read service time (r_await) of the agent host's disks, averaged across them and weighted by each disk's read IOPS (a plain average while the disks are idle). The agent polls /proc/diskstats every 5 seconds, computes the delta, and reports the aggregated value on each heartbeat. The figure is identical to what iostat -x shows in its r_await column.

This is a host-wide measure of how long each individual read I/O is taking, not a probe file read time and not specific to the agent's working directory. Any process on the host that issues reads contributes to the sample.

Platform support​

PlatformWhat is reported
LinuxAggregated r_await across all disks (ms). Real value.
Other / unsupported-1, displayed as "N/A".

Idle-disk behavior​

When neither read nor write IOPS were observed in the last 5-second sample window, the Live Agents disk-I/O pill shows "idle" in gray instead of "0 ms". The chart point for that sample will still be 0 — interpret it together with the IOPS pill (R/s, W/s) on the agent card to distinguish "fast disk" from "no I/O happened".

When the kernel reports 0 ms because the actual latency is below the integer millisecond resolution of /proc/diskstats (SSD/NVMe), the pill displays "< 1 ms" rather than literal zero.

Why it matters​

The copy engine writes per-page checkpoints, ClickHouse spool entries, and step logs to disk during every copy job. High read latency on the same backing device generally implies high write latency too, and will cause copy jobs to stall between pages. If the agent is running on network-attached or slow-backed storage, this chart will reveal the I/O pressure.

How to read it​

Value rangeInterpretation
0 / "< 1 ms"Sub-millisecond — local SSD or NVMe
1–5 msHealthy local disk or well-provisioned volume
5–20 msNormal for network-backed / shared storage
20–100 msElevated — storage contention or slow backing store
> 100 msProblematic — checkpoint writes will lag; consider moving to faster storage

Threshold coloring (Live Agents disk-I/O pill)​

Pill colorr_await range
Green< 5 ms
Amber5 – 19.99 ms
Red≥ 20 ms
GrayN/A (unsupported or idle)

Polling interval​

Sampled every 5 seconds inside the agent and reported on the heartbeat (~30 s + 0–10 s jitter). Spikes that last only one sample are usually transient I/O bursts. Sustained elevation indicates a persistent storage performance problem.

Tips​

  • Correlate read-latency spikes with copy job throughput drops — if objects/sec falls when latency spikes, the underlying device is the bottleneck.
  • Check the storage type where the agent writes its data/ directory. Local SSD/NVMe typically delivers < 1 ms; NFS-backed or network storage can spike to 50–200 ms under load.
  • On non-Linux hosts the value is intentionally "N/A".