one final word of warning for everyone.
while i no longer have the performance glitch....
I can no longer reproduce it.
Doing
ceph config set global rbd_cache true
does not seem to reproduce the old behaviour. even if i do things like unmap and remap the
test rbd.
Which is worrying. because if I cant control the behaviour.... who is to say it wont
mysteriously come back?
----- Original Message -----
From: "Philip Brown" <pbrown(a)medata.com>
To: "Sebastian Trojanowski" <sebcio.t(a)gazeta.pl>
Cc: "ceph-users" <ceph-users(a)ceph.io>
Sent: Thursday, December 17, 2020 9:02:05 AM
Subject: Re: [ceph-users] Re: performance degredation every 30 seconds
I am happy to say, this seems to have been the solution.
After running
ceph config set global rbd_cache false
I can now run the full 256 thread varient,
fio --direct=1 --rw=randwrite --bs=4k --ioengine=libaio --filename=/dev/rbd0
--iodepth=256 --numjobs=1 --time_based --group_reporting --name=iops-test-job
--runtime=120 --eta-newline=1
and there is no longer a noticeable performance dip.
Thanks Sebastian