Repository navigation
lvm: Deduplicate LVs by VG+LV name, not LV name alone - #1229
vojtechtrefny merged 1 commit into
Conversation
bd_lvm_lvs and bd_lvm_lvs_tree skipped multi-segment duplicates using only lv_name, which dropped same-named LVs from different VGs. Blivet then KeyErrors looking up thin pools such as vgold-tpool0 during Anaconda storage discovery. Co-authored-by: Cursor <cursoragent@cursor.com>
|
No actionable comments were generated in the recent review. Walkthrough
Merge Risk:
|
092f776
into
storaged-project:master
Summary
bd_lvm_lvs/bd_lvm_lvs_treecurrently treat multi-segmentlvsrows as duplicates using onlylv_name(from lvm: Do not include duplicate entries in bd_lvm_lvs output #671). That also drops legitimate same-named LVs that live in different VGs.bd_lvm_lvs(NULL), a second VG’s thin pool (e.g.vgold/tpool0vsvgnew/tpool0) can be omitted, and blivet then raisesKeyErrorlooking up the pool while resolving thin LVs.vg_nameandlv_namematch. Multi-segment duplicates within one LV are still collapsed.Test plan
test_lvs_same_name_different_vgs(BlockDev.lvm_lvs(None)must return both VGs’ LVs with the same name)reset()no longer KeyErrors)Related
Fixes the libblockdev side of RHEL Anaconda failures with shared LV names across VGs (RHEL-305266 / RHEL-305267).
Made with Cursor
Summary by CodeRabbit