@MarcThanks Marc,I am executing your profile via fio. I will send you the result. But what is consuming me is: tell bench sometimes return for example 2~10 and some times return 170~200.If the disk is burned out why sometimes return higher value?Currently this OSD is weighted 0. So there are no any load of ceph replication, deep scrubbing and so on.
Thanks.
On Saturday, December 24, 2022 at 06:57:05 PM GMT+3:30, Marc <marc@xxxxxxxxxxxxxxxxx> wrote:
>
> In my cluster, there are several OSDs of type ordinary SSD with very
> slow iops.
I think there have been several posts here about ordinary ssd's becoming slow under specific conditions. Why do you think your 'ordinary ssds' do not have this?
What does fio say about these disks?
I think this issue was related to how long these disks were under load, so maybe increase the runtime in this fio script.
[global]
ioengine=libaio
#ioengine=posixaio
invalidate=1
ramp_time=30
iodepth=1
runtime=180
time_based
direct=1
filename=/dev/sdX
#filename=/mnt/disk/fio-bench.img
[write-4k-seq]
stonewall
bs=4k
rw=write
[randwrite-4k-seq]
stonewall
bs=4k
rw=randwrite
fsync=1
[read-4k-seq]
stonewall
bs=4k
rw=read
[randread-4k-seq]
stonewall
bs=4k
rw=randread
fsync=1
[rw-4k-seq]
stonewall
bs=4k
rw=rw
[randrw-4k-seq]
stonewall
bs=4k
rw=randrw
[randrw-4k-d4-seq]
stonewall
bs=4k
rw=randrw
iodepth=4
[randread-4k-d32-seq]
stonewall
bs=4k
rw=randread
iodepth=32
[randwrite-4k-d32-seq]
stonewall
bs=4k
rw=randwrite
iodepth=32
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
