Bug #14117
closedKRBD : Performance issues with kernel 4.1 series
0%
Description
I am finding that it is a specific client that is experiencing this issue. Doing an RBD write from another client has no problem and OK throughput:
rbd map spin1/rbdtest
/dev/rbd0
root@lab2-mon1:/usr/local/ISS/supercore/temp/operations# dd if=/dev/zero of=/dev/rbd0 bs=1M count=1000
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 4.36296 s, 240 MB/s
Versus
root@e1:~# dd if=/dev/zero of=/dev/rbd0 bs=1M count=1000
1000+0 records in
1000+0 records out
1048576000 bytes (1.0 GB) copied, 46.3618 s, 22.6 MB/s
This problem looks like in kernel with 4.1 series :
kernel 4.1.6 is slow
kernel 4.1.15 is slow
As it is good with:
kernel 3.19 is fast
kernel 4.2.8 is fast
Updated by Ilya Dryomov over 8 years ago
- Project changed from rbd to Linux kernel client
- Assignee set to Ilya Dryomov
Alex posted on ceph-users and I replied there. Quoting:
"This is due to an old regression in blk-mq. rbd was switched to blk-mq
infrastructure in 4.0, the regression in blk-mq core was fixed in 4.2
by commit e6c4438ba7cb "blk-mq: fix plugging in blk_sq_make_request".
It's outside of rbd and wasn't backported, so we are kind of stuck with
it."
Updated by Vikhyat Umrao over 8 years ago
- Project changed from Linux kernel client to rbd
- Assignee deleted (
Ilya Dryomov)
Thanks Ilya for your feedback.
I am pasting the discussion : http://lists.ceph.com/pipermail/ceph-users-ceph.com/2015-December/006878.html
Updated by Vikhyat Umrao over 8 years ago
- Project changed from rbd to Linux kernel client
Updated by Ilya Dryomov over 8 years ago
- Status changed from New to Won't Fix
A bit more info on this: http://lists.ceph.com/pipermail/ceph-users-ceph.com/2015-December/006895.html, closing.