2

How do you define the Ceph OSD Disk Partition Size?

It always creates with only 10 GB usable space.

  • Disk size = 3.9 TB
  • Partition size = 3.7 TB
  • Using ceph-disk prepare and ceph-disk activate (See below)
  • OSD created but only with 10 GB, not 3.7 TB

.

Commands Used

root@proxmox:~# ceph-disk prepare --cluster ceph --cluster-uuid fea02667-f17d-44fd-a4c2-a8e19d05ed51 --fs-type xfs /dev/sda4

meta-data=/dev/sda4              isize=2048   agcount=4, agsize=249036799 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=1, sparse=0, rmapbt=0, reflink=0
data     =                       bsize=4096   blocks=996147194, imaxpct=5
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0 ftype=1
log      =internal log           bsize=4096   blocks=486399, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0

.

root@proxmox:~# ceph-disk activate /dev/sda4

creating /var/lib/ceph/tmp/mnt.jyqJTM/keyring
added entity osd.0 auth auth(auid = 18446744073709551615 key=AQAohgpdjwb3NRAAIrINUiXDWQ5iMWp4Ueah3Q== with 0 caps)
got monmap epoch 3
2019-06-19 19:59:54.006226 7f966e628e00 -1 bluestore(/var/lib/ceph/tmp/mnt.jyqJTM/block) _read_bdev_label failed to open /var/lib/ceph/tmp/mnt.jyqJTM/block: (2) No such file or directory
2019-06-19 19:59:54.006285 7f966e628e00 -1 bluestore(/var/lib/ceph/tmp/mnt.jyqJTM/block) _read_bdev_label failed to open /var/lib/ceph/tmp/mnt.jyqJTM/block: (2) No such file or directory
2019-06-19 19:59:55.668619 7f966e628e00 -1 created object store /var/lib/ceph/tmp/mnt.jyqJTM for osd.0 fsid fea02667-f17d-44fd-a4c2-a8e19d05ed51
Created symlink /run/systemd/system/ceph-osd.target.wants/ceph-osd@0.service → /lib/systemd/system/ceph-osd@.service.

# Don't worry about my keys/IDs, its just a dev environment.

.

Disk Layout

root@proxmox:~# fdisk -l
Disk /dev/sda: 3.9 TiB, 4294967296000 bytes, 8388608000 sectors
OMITTIED

Device         Start        End    Sectors   Size Type
/dev/sda1         34       2047       2014  1007K BIOS boot
/dev/sda2       2048    1050623    1048576   512M EFI System
/dev/sda3    1050624  419430400  418379777 199.5G Linux LVM
/dev/sda4  419430408 8388607966 7969177559   3.7T Ceph OSD

.

Ceph OSD Disk Size Incorrect (10 GB not 1.7 TB)

root@proxmox:~# ceph status
  data:
    pools:   0 pools, 0 pgs
    objects: 0 objects, 0B
    usage:   1.00GiB used, 9.00GiB / 10GiB avail
    pgs:

.

.

Full Install Details

If you want details on Proxmox install and creating Ceph OSD with partitions, read on...

Setup

  • Disk Size: 2TB NVMe (/dev/sda)
  • Operating system (Proxmox) installed with 200 GB and rest of disk is empty (1800 GB).
  • Once booted and in the web interface, create a cluster and join two hosts to ensure a green quorum status
  • Now do script below

Config Script

# Install Ceph
pveceph install

# Configure Network (Just run on Primary Proxmox Server, your LAN network)
pveceph init --network 192.168.6.0/24

# Create Monitor
pveceph createmon

# View Disks Before
sgdisk --print /dev/sda

sgdisk --largest-new=4 --change-name="4:CephOSD" \
--partition-guid=4:4fbd7e29-9d25-41b8-afd0-062c0ceff05d \
--typecode=4:4fbd7e29-9d25-41b8-afd0-062c0ceff05d -- /dev/sda
  
# View Disks After (Compare)
sgdisk --print /dev/sda

# Reboot for changes to be in affect
reboot

# Note your cluster ID (fsid) at this point from the web interface 
Datacenter > Server > Ceph

# Prepare the Ceph OSD Disk, replace cluster-uuid with above fsid
ceph-disk prepare --cluster ceph --cluster-uuid fea02667-f17d-44fd-a4c2-a8e19d05ed51 --fs-type xfs /dev/sda4

# Activate the Ceph OSD Disk 
ceph-disk activate /dev/sda4

# Check Ceph OSD Disk Size
ceph status

Notes I have read posts highly recommending to use disks instead of partitions because of performance issues, I understand the warnings, but In my case I'm using NVMe SSD storage and accept any risks.

Community
  • 1
rwcommand
  • 143
  • 6

1 Answers1

0

We ran into a similar issue recently where our OSD was limited to only 10GB despite a much larger actual disk space and found that it's related to the bluestore backend in Ceph. When bluestore is backed by a file, if that file does not already exist, it will create it:

bluestore(/var/lib/ceph/osd/ceph-0/block) _read_bdev_label failed to open /var/lib/ceph/osd/ceph-0/block: (2) No such file or directory
bluestore(/var/lib/ceph/osd/ceph-0) mkfs generated fsid db688b70-0f63-4086-ba4a-ef36dc4a3147

and then resize it based on the bluestore_block_size configuration:

bluestore(/var/lib/ceph/osd/ceph-0) _setup_block_symlink_or_file resized block file to 10 GiB

The default value for bluestore_block_size in ceph version 14.2.2, for example, is 10GB [1]:

    Option("bluestore_block_size", Option::TYPE_SIZE, Option::LEVEL_DEV)
    .set_default(10_G)
    .set_flag(Option::FLAG_CREATE)
    .set_description("Size of file to create for backing bluestore"),

This can be adjusted by setting the following in your ceph.conf:

[global]
bluestore_block_size = <block file size in bytes>

[1] https://github.com/ceph/ceph/blob/v14.2.2/src/common/options.cc#L4338-L4341

melwitt
  • 1