Questions tagged [fio]

The tag has no usage guidance.

Filter by
Sorted by
Tagged with
9 votes
1 answer
463 views

Understanding ZFS NVMe benchmarks with FIO

I'm trying to benchmark an all-NVMe ZFS disk array. I'm familiar with extremely fast benchmark results, with very little disk activity due to efficient ZFS caching. I ran into the reverse: massive ...
Frank Vermeulen's user avatar
4 votes
1 answer
185 views

why read is faster when using O_DIRECT flag?

I copied a 10GB of file in my SSD which has read bandwidth of around 3.3GB/s, benchmarked using fio command. Here is the ref: https://cloud.google.com/compute/docs/disks/benchmarking-pd-performance I ...
Abdul Wadood's user avatar
0 votes
0 answers
174 views

ZoL+FIO Randwrite nvme bs=8k@32io = 148MiB/s?

Sorry for the title, but it's very short summary of BS that I'm looking into. Situation I'm using ZoL 2.1.5 (from jonathonf's ppa) on Ubuntu (tried 20.04&22.04) I have following NVMe disk ...
Vladislav Losev's user avatar
0 votes
2 answers
94 views

System auto restart during the FIO stress test on a RAID0 array(6x SSDs) created by mdadm under Yocto OS

I met a problem when executing FIO stress test on a RAID0 which built from 6x SSDs thru mdadm under Yocto OS, below is the information: 6x PCIe NVMe SSD are the same vendor and model which is with 1....
Jacky Lee's user avatar
1 vote
0 answers
421 views

FIO test results wildly vary based on number of jobs

Take the following 3 commands: fio --name=write_throughput --numjobs=8 \ --size=10G --time_based --runtime=60s --ramp_time=2s --ioengine=libaio \ --direct=1 --verify=0 --bs=1M --iodepth=64 --rw=write \...
KdgDev's user avatar
  • 273
0 votes
1 answer
887 views

How to read and decode fio --bandwidth-log?

I'm Looking for assistance with reading and decoding fio --bandwidth-log. I've run the below command and the output includes a few columns as listed below, how to read and decode each column? fio --...
Or Lapid's user avatar
2 votes
1 answer
461 views

Interpreting fio test result

I ran fio on a VM provided by my VPS provider and got this result: $ fio --name=random-write --ioengine=posixaio --rw=randwrite --bs=4k --numjobs=1 --size=4g --iodepth=1 --runtime=60 --time_based --...
pepoluan's user avatar
  • 5,038
0 votes
0 answers
1k views

FIO 17x slower on ZFS than raw device

I have ZFS pool RAIDZ-1 on 4 x 3.84TB SATA3 'enterprise' SSDs. zfs create rpool/fio zfs set primarycache=none rpool/fio fio --ioengine=sync --filename=[X] --direct=1 --rw=read --bs=4K --numjobs=1 --...
simonw's user avatar
  • 35
0 votes
1 answer
313 views

How to correctly benchmark sequential read speeds on 2.5" hard drive with fio on FreeBSD?

Like suggested in https://askubuntu.com/a/991311 I tested the read speed of a 2.5" spinning hard drive on a TrueNAS/FreeBSD system. For "Sequential READ speed with big blocks": fio --...
Andept's user avatar
  • 73
-1 votes
1 answer
403 views

Simulate some process IO with fio or other tool

Is there any way to track a specific process and get stats for disk IO such as queue depth, total reading/writing threads, percent of reads/writes and so on? The main goal is to use all that ...
Maxim's user avatar
  • 103
1 vote
1 answer
397 views

`fio` does not timeout even specified?

# fio --name=random-write --directory=/mnt/test/ --ioengine=posixaio --rw=randwrite -bs=4k --numjobs=1 --size=4g -iodepth=1 -runtime=600 --time_based --end_fsync=1 random-write: (g=0): rw=randwrite, ...
HCSF's user avatar
  • 245
2 votes
2 answers
310 views

AWS XFS performance issues. Stripped setup vs single disk setup

I wanted to share an issue (I guess I may be misunderstanding some concepts) that I'm facing with some benchmarks I'm doing to XFS setups, as we are going to recently migrate a service to a new ...
David Barranco's user avatar
6 votes
1 answer
721 views

fio 3.23 core dumps when bench-marking many small files

I have been asked to come up fio benchmark results for this test dataset: 1048576x1MiB. So, overall size is 1TiB. The set contains 2^20 1MiB files. The server runs CentOS Linux release 7.8.2003 (Core)...
foss4me's user avatar
  • 63
7 votes
0 answers
2k views

Write performance is 5 times worse with LUKS on top of mdadm RAID10 than without LUKS

I have servers with many NVMe disks. I am testing disk performance with fio using the following: fio --name=asdf --rw=randwrite --direct=1 --ioengine=libaio --bs=16k --numjobs=8 --size=10G --runtime=...
tacos_tacos_tacos's user avatar