"bluestore-block-db-size" config doesn't accept a relative value like 100%

Bug #1905026 reported by Nobuto Murata
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Ceph OSD Charm
New
Undecided
Unassigned

Bug Description

When specifying a separate and faster drive for "bluestore-db", it would make sense to set bluestore-block-db-size config as "please use the entire drive". However, the current charm needs an exact value in bytes which is not ideal as we never know the exact bytes until we actually deploy a machine.

https://jaas.ai/ceph-osd#charm-config-bluestore-block-db-size

> bluestore-block-db-size
> (int) Size (in bytes) of a partition, file or LV to use for BlueStore metadata or RocksDB SSTs, provided on a per backend device basis. . Example: 128 GB device, 8 data devices provided in "osd-devices" gives 128 / 8 GB = 16 GB = 16000000000 bytes per device. . A default value is not set as it is calculated by ceph-disk (before Luminous) or the charm itself, when ceph-volume is used (Luminous and above).

Revision history for this message
Paul Goins (vultaire) wrote :

While not an exact duplicate, there is a relation between this and https://bugs.launchpad.net/charm-ceph-osd/+bug/1827463, which argues for a percentage-based default size.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.