Tune all bcache devices on a unit with best-practice tuning options.
This charm performs best-practice performance tuning of bcache devices on SD/NVMe based installations.
Disable performance based congestion behaviour - the SSD or NVMe will in all likelyhood always be faster than the underlying spindle:
congested_read_threshold_us: 0 congested_write_threshold_us: 0
Disable writethrough of sequential data writes:
Increase the amount of dirty data the cache device will hold before starting to persist to the backing device:
To use this charm, simple deploy and relation to principle services:
juju deploy bcache-tuning juju add-relation bcache-tuning ceph-osd juju add-relation bcache-tuning nova-compute
This charm presents no configuration options.
- (boolean) Enable debug logging.