If you use block_db_size and limit in your yaml file, e.g. block_db_size: 64G (or whatever you choose) limit: 6 this should not consume the entire disk but only as much as you configured. Can you try if that works for you? Zitat von "Schweiss, Chip" <chip@innovates.com>:
I'm trying to set up a new ceph cluster with cephadm on a SUSE SES trial that has Ceph 15.2.8
Each OSD node has 18 rotational SAS disks, 4 NVMe 2TB SSDs for DB, and 2 NVME2 200GB Optane SSDs for WAL.
These servers will eventually have 24 rotational SAS disks that they will inherit from existing storage servers. So I don't want all the space used on the DB and WAL SSDs.
I suspect from the comment "(db_slots is actually to be favoured here, but it's not implemented yet)" on this page in the docs: https://docs.ceph.com/en/latest/cephadm/drivegroups/#the-advanced-case these parameters are not yet implemented, yet are documented as such under "ADDITIONAL OPTIONS"
My osd_spec.yml: service_type: osd service_id: three_tier_osd placement: host_pattern: '*' data_devices: rotational: 1 model: 'ST14000NM0288' db_devices: rotational: 0 model: 'INTEL SSDPE2KX020T8' limit: 6 wal_devices: model: 'INTEL SSDPEL1K200GA' limit: 12 db_slots: 6 wal_slots: 12
All available space is consumed on my DB and WAL SSDs with only 18 OSDs, leaving no room to add additional spindles.
Is this still work in progress, or a bug I should report? Possibly related to https://github.com/rook/rook/issues/5026 At the minimum, this appears to be a documentation bug.
How can I work around this?
-Chip _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io