Re: blk-mq 5-8 times slower for bmap-tools
From: Paolo Valente <hidden>
Date: 2018-08-17 20:46:36
Il giorno 17 ago 2018, alle ore 19:39, Ricardo Ribalda Delgado =
[off-list ref] ha scritto:
=20 Hello Paolo On Fri, Aug 17, 2018 at 7:35 PM Paolo Valente =
[off-list ref] wrote:
quoted
=20 =20 =20quoted
Il giorno 17 ago 2018, alle ore 19:31, Ricardo Ribalda Delgado =
[off-list ref] ha scritto:
quoted
quoted
=20 Hello =20 Coming from: =
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=3D906328
quoted
quoted
=20 bmaptools is kind of a "smart dd" tool, that lets you write images very fast and securely. Since the last Debian Kernel update it has become 5-8 times slower. =20 After some debugging, we have figured out that the reason for that slowness is the Multi-Queue Block IO Queueing Mechanism. =20 Debian maintainer has pointed out that in the near future the single queue will be deprecated. =20 My question is if we can get a similar perfomance for bmaptools with blk-mq and how? =20=20 Have you also checked what happens after switching to a different I/O scheduler for the involved drive (among none, mq-deadline, bfq and kyber)?=20 I have only tried mq-deadline and none (because they are enabled by default on Debian). Both produce results in the same range: 5-8 times slower. =20 I could easily enable kyber: cat /boot/config-4.17.0-1-amd64 | grep CONFIG_MQ_IOSCHED CONFIG_MQ_IOSCHED_DEADLINE=3Dy CONFIG_MQ_IOSCHED_KYBER=3Dm =20 But I left the card reader on the office, so any test would have to wait until monday sorry :( =20
Try bfq too, as it is probably the one containing also the most old way heuristics. If bfq fails too, we will at least know that the issue is out of the schedulers. Or in every scheduler! :) Paolo
=20quoted
=20 Paolo =20quoted
Thanks! =20 -- Ricardo Ribalda=20=20 =20 --=20 Ricardo Ribalda