Reclaim metadata when arc_available_memory < 0 even if meta_used is n…

…ot bigger than arc_meta_limit As described in openzfs#14054 if zfs_arc_meta_limit_percent=100 then ARC target can collapse to arc_min due to arc_purge not freeing any metadata. This patch lets arc_prune to do its work when arc_available_memory is negative even if meta_used is not bigger than arc_meta_limit, avoiding ARC target collapse. Signed-off-by: Gionatan Danti <[email protected]>
shodanshok · Oct 30, 2022 · 2a1503b · 2a1503b
1 parent 41fc313
commit 2a1503b
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/module/zfs/arc.c b/module/zfs/arc.c
@@ -4469,7 +4469,7 @@ arc_evict_meta_balanced(uint64_t meta_used)
 	 * meta buffers.  Requests to the upper layers will be made with
 	 * increasingly large scan sizes until the ARC is below the limit.
 	 */
-	if (meta_used > arc_meta_limit) {
+	if (meta_used > arc_meta_limit || arc_available_memory() < 0) {
 		if (type == ARC_BUFC_DATA) {
 			type = ARC_BUFC_METADATA;
 		} else {