On Sat, 15 Feb 2014, Samuel Terburg - Panther-IT BV wrote:
I have a performance problem i would like advise.
I have the following sub-optimal setup:
* 2 Servers (WTFM008 WTFM009)
* HP Proliant DL180
* SmartArray G6 P410 raid-controller
* 4x 500GB RAID5 (seq writes = 230MB/s)
* CentOS 6.5 x86_64
* 2.000.000 files (ms-word), with no directory structure
* Ceph
* ceph-deploy mon create WTFM008 WTFM009
* ceph-deploy mds create WTFM008 WTFM009
* ceph-deploy osd activate WTFM008:/var/lib/ceph/osd/ceph-0
WTFM009:/var/lib/ceph/osd/ceph-1
(osd is using root fs)
* ceph-fuse /mnt/ceph
I am currently trying to copy 2 million ms-word documents into ceph.
When i started it was doing about 10 files per second.
Now, 1 week later, it has done about 500.000 files and has slowed down to 1
file per 10 seconds.
How can i improve this terrible performance?
You probably need to add
mds frag = true
in the [mds] section
sage
* The hardware is a fixed configuration, i cannot add (SSD) disks or change
RAID.
* I could not find the cephfs kernel module so i had to use cephfs-fuse.
* I could have started with a degraded setup (1 OSD) for the initial load,
would that have helped in the performance? (Ceph not having to do the
distribution part)
* There is nu load on the systems at all (not cpu, not mem, not disk i/o)
Below is my crush map.
Regards,
Samuel Terburg
Panther-IT BV
# begin crush map
# devices
device 0 osd.0
device 1 osd.1
# types
type 0 osd
type 1 host
type 2 rack
type 3 row
type 4 room
type 5 datacenter
type 6 root
# buckets
host WTFM008 {
id -2 # do not change unnecessarily
# weight 1.340
alg straw
hash 0 # rjenkins1
item osd.0 weight 1.340
}
host WTFM009 {
id -3 # do not change unnecessarily
# weight 1.340
alg straw
hash 0 # rjenkins1
item osd.1 weight 1.340
}
root default {
id -1 # do not change unnecessarily
# weight 2.680
alg straw
hash 0 # rjenkins1
item WTFM008 weight 1.340
item WTFM009 weight 1.340
}
# rules
rule data {
ruleset 0
type replicated
min_size 1
max_size 10
step take default
step chooseleaf firstn 0 type host
step emit
}
rule metadata {
ruleset 1
type replicated
min_size 1
max_size 10
step take default
step chooseleaf firstn 0 type host
step emit
}
rule rbd {
ruleset 2
type replicated
min_size 1
max_size 10
step take default
step chooseleaf firstn 0 type host
step emit
}
# end crush map
# ceph -w
cluster 4f7bcb26-0cee-4472-abca-c200a999b686
health HEALTH_OK
monmap e1: 2 mons at
{WTFM008=192.168.0.1:6789/0,WTFM009=192.168.0.2:6789/0}, election epoch 4,
quorum 0,1 WTFM008,WTFM009
mdsmap e5: 1/1/1 up {0=WTFM008=up:active}, 1 up:standby
osdmap e14: 2 osds: 2 up, 2 in
pgmap v151668: 192 pgs, 3 pools, 31616 MB data, 956 kobjects
913 GB used, 1686 GB / 2738 GB avail
192 active+clean
client io 40892 kB/s rd, 7370 B/s wr, 1 op/s
___
ceph-users mailing list
ceph-users@lists.ceph.com
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com