docker-diskmark

A fio-based disk benchmark docker container, similar to what CrystalDiskMark does.

Basic Usage

# From Docker Hub
docker pull e7db/diskmark
docker run -it --rm e7db/diskmark

# From GitHub Container Registry
docker pull ghcr.io/e7db/diskmark
docker run -it --rm ghcr.io/e7db/diskmark

Options

All options can be configured via CLI arguments (recommended) or environment variables. CLI arguments take precedence.

The container supports multiple CLI formats: --key value, --key=value, and -k value.

Short	Long	Env Var	Default	Description
-h	--help			Show help message and exit
-v	--version			Show version information and exit
-t	--target	TARGET	/disk	Target directory for benchmark
-p	--profile	PROFILE	auto	Benchmark profile: auto, default, nvme
-j	--job	JOB		Custom job definition (e.g., RND4KQ32T16). Overrides profile.
-i	--io	IO	direct	I/O mode: direct (sync), buffered (async)
-d	--data	DATA	random	Data pattern: random, zero
-s	--size	SIZE	1G	Test file size (e.g., 500M, 1G, 10G)
-w	--warmup	WARMUP=1	1	Enable warmup phase
	--no-warmup	WARMUP=0		Disable warmup phase
	--warmup-size	WARMUP_SIZE	(profile)	Warmup block size (8M default, 64M nvme)
-r	--runtime	RUNTIME	5s	Runtime per job (e.g., 500ms, 5s, 2m)
-l	--loops	LOOPS		Number of test loops
-n	--dry-run	DRY_RUN=1	0	Validate configuration without running
-f	--format	FORMAT		Output format: json, yaml, xml
-u	--no-update-check	UPDATE_CHECK=0	1	Disable update check at startup
	--color	COLOR=1		Force colored output
	--no-color	COLOR=0		Disable colored output
	--emoji	EMOJI=1		Force emoji output
	--no-emoji	EMOJI=0		Disable emoji output

Profiles

The container contains two different test profiles:

• Default profile:
  • Sequential 1M Q8T1
  • Sequential 1M Q1T1
  • Random 4K Q32T1
  • Random 4K Q1T1
• NVMe profile:
  • Sequential 1M Q8T1
  • Sequential 128K Q32T1
  • Random 4K Q32T16
  • Random 4K Q1T1

Examples

Basic parameters

# 4 GB file, 2 loops, warmup, zero data pattern
docker run -it --rm e7db/diskmark --size 4G --warmup --loops 2 --data zero
docker run -it --rm e7db/diskmark -s 4G -w -l 2 -d zero

# Hybrid mode: 3 loops, each capped at 10 seconds
docker run -it --rm e7db/diskmark --loops 3 --runtime 10s

# Custom warmup block size
docker run -it --rm e7db/diskmark --warmup --warmup-size 128M

Force profile

Drive detection selects the appropriate profile (default or nvme). Override if needed:

docker run -it --rm e7db/diskmark --profile nvme

Custom job

Run a custom job using the format [RND|SEQ][size][Q depth][T threads]:

• RND or SEQ — random or sequential access
• xxK or xxM — block size (e.g., 4K, 1M)
• Qyy — queue depth
• Tzz — number of threads

Example: RND4KQ32T16 = random 4K blocks, queue depth 32, 16 threads.

docker run -it --rm e7db/diskmark --job RND4KQ32T16

Specific disk

By default, the benchmark uses a Docker volume at /disk. Mount a different path to benchmark another disk:

docker run -it --rm -v /path/to/disk:/disk e7db/diskmark

Machine-readable output

Output in JSON, YAML, or XML for scripting (automatically disables colors, emojis, and update check):

docker run -it --rm e7db/diskmark --format json
docker run -it --rm e7db/diskmark --format yaml
docker run -it --rm e7db/diskmark --format xml

JSON output sample

{
  "configuration": {
    "target": "/mnt",
    "drive": {
      "label": "NVMe",
      "info": "Samsung SSD 990 PRO 2TB"
    },
    "filesystem": {
      "type": "ext4",
      "partition": "/dev/nvme0n1p1",
      "size": "1.8T"
    },
    "profile": "nvme",
    "io": "direct",
    "data": "random",
    "size": "1G",
    "warmup": 1,
    "runtime": "5s"
  },
  "results": [
    {"name": "SEQ1MQ8T1", "status": "success", "read": {"bandwidth_mb": 524, "iops": 499, "latency_ms": 2.15}, "write": {"bandwidth_mb": 498, "iops": 475, "latency_ms": 1.89}},
    {"name": "SEQ1MQ1T1", "status": "success", "read": {"bandwidth_mb": 512, "iops": 488, "latency_ms": 0.52}, "write": {"bandwidth_mb": 487, "iops": 464, "latency_ms": 0.48}},
    {"name": "RND4KQ32T1", "status": "success", "read": {"bandwidth_mb": 45, "iops": 11691, "latency_ms": 2.73}, "write": {"bandwidth_mb": 42, "iops": 10839, "latency_ms": 2.95}},
    {"name": "RND4KQ1T1", "status": "success", "read": {"bandwidth_mb": 41, "iops": 10583, "latency_ms": 0.09}, "write": {"bandwidth_mb": 38, "iops": 9814, "latency_ms": 0.10}}
  ]
}

YAML output sample

configuration:
  target: "/mnt"
  drive:
    label: "NVMe"
    info: "Samsung SSD 990 PRO 2TB"
  filesystem:
    type: "ext4"
    partition: "/dev/nvme0n1p1"
    size: "1.8T"
  profile: "nvme"
  io: "direct"
  data: "random"
  size: "1G"
  warmup: 1
  runtime: "5s"
results:
  - name: "SEQ1MQ8T1"
    status: "success"
    read:
      bandwidth_mb: 524
      iops: 499
      latency_ms: 2.15
    write:
      bandwidth_mb: 498
      iops: 475
      latency_ms: 1.89
  - name: "SEQ1MQ1T1"
    status: "success"
    read:
      bandwidth_mb: 512
      iops: 488
      latency_ms: 0.52
    write:
      bandwidth_mb: 487
      iops: 464
      latency_ms: 0.48
  - name: "RND4KQ32T1"
    status: "success"
    read:
      bandwidth_mb: 45
      iops: 11691
      latency_ms: 2.73
    write:
      bandwidth_mb: 42
      iops: 10839
      latency_ms: 2.95
  - name: "RND4KQ1T1"
    status: "success"
    read:
      bandwidth_mb: 41
      iops: 10583
      latency_ms: 0.09
    write:
      bandwidth_mb: 38
      iops: 9814
      latency_ms: 0.10

XML output sample

<?xml version="1.0" encoding="UTF-8"?>
<benchmark>
  <configuration>
    <target>/mnt</target>
    <drive label="NVMe">Samsung SSD 990 PRO 2TB</drive>
    <filesystem type="ext4" partition="/dev/nvme0n1p1" size="1.8T" />
    <profile>nvme</profile>
    <io>direct</io>
    <data>random</data>
    <size>1G</size>
    <warmup>1</warmup>
    <runtime>5s</runtime>
  </configuration>
  <results>
    <job name="SEQ1MQ8T1" status="success">
      <read bandwidth_mb="524" iops="499" latency_ms="2.15" />
      <write bandwidth_mb="498" iops="475" latency_ms="1.89" />
    </job>
    <job name="SEQ1MQ1T1" status="success">
      <read bandwidth_mb="512" iops="488" latency_ms="0.52" />
      <write bandwidth_mb="487" iops="464" latency_ms="0.48" />
    </job>
    <job name="RND4KQ32T1" status="success">
      <read bandwidth_mb="45" iops="11691" latency_ms="2.73" />
      <write bandwidth_mb="42" iops="10839" latency_ms="2.95" />
    </job>
    <job name="RND4KQ1T1" status="success">
      <read bandwidth_mb="41" iops="10583" latency_ms="0.09" />
      <write bandwidth_mb="38" iops="9814" latency_ms="0.10" />
    </job>
  </results>
</benchmark>