On Fri, Feb 05, 2021 at 09:03:28AM +0100, BERTRAND Joël wrote:
> LUN0 Storage /dev/dk6 Auto
>
> I can change dk6 by rdk6, but it's not a solution to fix issue with
> iscsid, when NetBSD initiator tries to use a iSCSI target. In my case,
> istgt runs as expected even if I can increase its performance.
If you use the block device as the iSCSI backend, the data is passed
through the buffer cache of the host. That's slow for the first
time, fast for repetitions (since the data is cached) unless it's
more data than fits into the buffer cache, then it's ugly slow.
vm.bufcache is 15 (percent) by default. On your 16GB machine
that means that I/O of up to about 2GB may look fine, but
if you move more, things will crawl.
You also get the effect that writing seems to be fine at the start
as data is only buffered and slowly written out in the background.
Once the cache fills or isn't yet flushed from a previous run, speed
drops to very low levels.
But that doesn't explain anything with your qNAP NAS.
Greetings,
--
Michael van Elst
Internet: mlel...@serpens.de
"A potential Snark may lurk in every tree."
