[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ceph-users
Subject:    [ceph-users] SSD partitioned for HDD wal+db plus SSD osd
From:       Chris Dunlop <chris () onthe ! net ! au>
Date:       2021-09-29 6:23:57
Message-ID: 20210929062357.GA2038445 () onthe ! net ! au
[Download RAW message or body]

Hi,

Is there any way of using "ceph orch apply osd" to partition an SSD as 
wal+db for a HDD OSD, with the rest of the SSD as a separate OSD?

E.g. on a machine (here called 'k1') with a small boot drive and a single 
HDD and SSD, this will create an OSD on the HDD, with wal+db on a 60G 
logical volume on the SSD:

$ ceph orch apply osd -i <(cat <<'END'
service_type: osd
service_id: k_hdd_ssd
service_name: osd.hdd_ssd_mix
placement:
   host_pattern: 'k1'
data_devices:
   rotational: 1
   size: '8T:'
db_devices:
   rotational: 0
filter_logic: AND
block_db_size: 60G
END
)

And k1 ends up with LVs like so, with free space on the SSD:

k1# pvs
   PV         VG                Fmt  Attr PSize    PFree
   /dev/sda   ceph-3842b9ec-... lvm2 a--  <223.57g <163.57g
   /dev/sdb   ceph-6ef9d5b3-... lvm2 a--    14.55t       0

Any suggestions on how to get "ceph orch apply" to use the remainder of 
the space on the SSD for another OSD, or is this something I'm going to 
have to create by hand?

In case you're wondering why...

This is for initially just for throughput testing using the in-place 
hardware. The HDD OSDs are for the data part of an erasure pool, and 
replicated partial-SSD OSDs for the metadata part, to see if the HDD 
wal+db on SSD makes any significant difference over pure-HDD OSDs for 
"mostly large writes" on rbd.  

Assuming it does make a difference, is there any particular reason to 
not go this way?

Chris
_______________________________________________
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-leave@ceph.io
[prev in list] [next in list] [prev in thread] [next in thread]