Hi,
I was trying to determine performance impact of deep-scrubbing with
osd_disk_thread_ioprio_class option set but it looks like it's ignored.
Performance (during deep-scrub) is the same with this options set or
left with defaults (1/3 of "normal" performance).
# ceph --admin-daemon /var/run/ceph/ceph-osd.26.asok config show | grep
osd_disk_thread_ioprio
"osd_disk_thread_ioprio_class": "idle",
"osd_disk_thread_ioprio_priority": "7",
# ps -efL | grep 'ce[p]h-osd --cluster=ceph -i 26' | awk '{ print $4; }'
| xargs --no-run-if-empty ionice -p | sort | uniq -c
18 unknown: prio 0
186 unknown: prio 4
# cat /sys/class/block/sdf/queue/scheduler
noop deadline [cfq]
And finallyGDB:
Breakpoint 1, ceph_ioprio_string_to_class (s=...) at
common/io_priority.cc:48
warning: Source file is more recent than executable.
48 return IOPRIO_CLASS_IDLE;
(gdb) cont
Continuing.
Breakpoint 2, OSD::set_disk_tp_priority (this=0x3398000) at osd/OSD.cc:8548
warning: Source file is more recent than executable.
8548 disk_tp.set_ioprio(cls,
cct->_conf->osd_disk_thread_ioprio_priority);
(gdb) print cls
$1 = -22
So the IO priorities are *NOT*set (cls >= 0). I'm not sure where this
-22 came from.Any ideas?
In the mean time I'll compile ceph from sources and check again.
Ceph installed from Ceph repositories:
# ceph-osd -v
ceph version 0.86 (97dcc0539dfa7dac3de74852305d51580b7b1f82)
# apt-cache policy ceph
ceph:
Installed: 0.86-1precise
Candidate: 0.86-1precise
Version table:
*** 0.86-1precise 0
500 http://eu.ceph.com/debian-giant/ precise/main amd64 Packages
100 /var/lib/dpkg/status
--
PS
_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
