Description of problem:
thin-provisioned logical volumes are in danger of exhausting metadata space when the underlying hardware is changed. For example, when new storage is attached and the number of PV extents increases for the thin pool. The danger is not easily perceived when looking at the output of "lvs --all". I'm suggesting that the output of "lvs --all" be ammended to include two additional fields: metadata_percent and lv_metadata_size
Actual results:
[jcall@jcall-laptop ~]$ sudo lvs --all
LV VG Attr LSize Pool Origin Data% Move Log Cpy%Sync Convert
home rootvg Vwi-aotz-- 97.66g pool00 2.25
images rootvg Vwi-aotz-- 488.28g pool00 13.52
[lvol0_pmspare] rootvg ewi------- 80.00m
pool00 rootvg twi-a-tz-- 634.77g 11.61
[pool00_tdata] rootvg Twi-ao---- 634.77g
[pool00_tmeta] rootvg ewi-ao---- 80.00m
root rootvg Vwi-aotz-- 48.83g pool00 11.24
swap rootvg -wi-ao---- 7.81g
Expected results:
[jcall@jcall-laptop ~]$ sudo lvs --all -o lv_name,vg_name,lv_attr,lv_size,pool_lv,origin,data_percent,metadata_percent,lv_metadata_size,move_pv,mirror_log,copy_percent,convert_lv
LV VG Attr LSize Pool Origin Data% Meta% MSize Move Log Cpy%Sync Convert
home rootvg Vwi-aotz-- 97.66g pool00 2.25
images rootvg Vwi-aotz-- 488.28g pool00 13.52
[lvol0_pmspare] rootvg ewi------- 80.00m
pool00 rootvg twi-a-tz-- 634.77g 11.61 6.25 80.00m
[pool00_tdata] rootvg Twi-ao---- 634.77g
[pool00_tmeta] rootvg ewi-ao---- 80.00m
root rootvg Vwi-aotz-- 48.83g pool00 11.24
swap rootvg -wi-ao---- 7.81g
Additional info:
I had an unfortunate incident that resulted in filesystem corruption due to this. Not being able to immediately see the utilization / allocated space of the metadata LV compounded the frustration. Additional details were posted at https://ask.fedoraproject.org/en/question/46400/why-dont-thin-provisoned-pools-alert-users-when-they-fill-up/