Hello, Am 08.03.19 um 20:53 schrieb linux-raid-20190308@xxxxxxxxxxxxxxx: since Linux 5.0 my md raid5 is painfully slow, ~100 kb/s, bisected that, bisect blames 5d2ee7122c73be6a3b6bfe90d237e8aed737cfaa is the first bad commit commit 5d2ee7122c73be6a3b6bfe90d237e8aed737cfaa Author: Jens Axboe <axboe@xxxxxxxxx> Date: Thu Nov 29 17:36:41 2018 -0700 when trying to revert that from v5.0 I get a conflict and do not know how to resolve that. Please help/fix. Best Regards, Michael Leun > Hello, > > I have a raid5 array consisting of 4 8TB disks connected via USB3. > > [ml@xenia ~]$ cat /proc/mdstat > Personalities : [raid6] [raid5] [raid4] > md127 : active raid5 sde1[4] sdc1[1] sdd1[2] sdb1[0] > 23441682432 blocks super 1.2 level 5, 512k chunk, algorithm 2 > [4/4] [UUUU] > bitmap: 0/59 pages [0KB], 65536KB chunk > > unused devices: <none> > [ml@xenia ~]$ > > Up to and including kernel 4.20.13-arch1 everything is fine and I get > reasonable performance: > > [root@xenia ml]# dd if=/dev/md127 of=/dev/null bs=2M status=progress > 3078619136 bytes (3,1 GB, 2,9 GiB) copied, 13 s, 237 MB/s^C > 1516+0 Datensätze ein > 1515+0 Datensätze aus > 3177185280 bytes (3,2 GB, 3,0 GiB) copied, 13,4403 s, 236 MB/s > > [root@xenia ml]# uname -a > Linux xenia.leun.net 4.20.13-arch1-1-ARCH #1 SMP PREEMPT Wed Feb 27 > 19:10:28 UTC 2019 x86_64 GNU/Linux > [root@xenia ml]# > > > With kernel 5.0.0-arch1-1-ARCH (also verified that with kernel.org > 5.0.0) I get PAINFULLY slow performance on the raid, but the performance > of the disks the array consists of is ok: > > [root@xenia ml]# dd if=/dev/md127 of=/dev/null bs=2M status=progress > 6291456 bytes (6,3 MB, 6,0 MiB) copied, 41 s, 154 kB/s^C > 4+0 Datensätze ein > 3+0 Datensätze aus > 6291456 bytes (6,3 MB, 6,0 MiB) copied, 59,0605 s, 107 kB/s > > [root@xenia ml]# dd if=/dev/sdb1 of=/dev/null bs=2M status=progress > 1642070016 bytes (1,6 GB, 1,5 GiB) copied, 8 s, 205 MB/s^C > 806+0 Datensätze ein > 805+0 Datensätze aus > 1688207360 bytes (1,7 GB, 1,6 GiB) copied, 8,23402 s, 205 MB/s > > [root@xenia ml]# > > (The other 3 disks show the same speed) > > Any idea why it is THAT slow? > > Best Regards, > > Michael Leun >
