a 2.4.27 box with linux-2.4.27-devmapper-ioctl.patch patch got a OOM several days ago. one workmate here claimed to only create several snapshots on a 100GB LV. unfortunately, i could not reproduce the error. fortunately i have the system monitor running on the box and got the slabinfo file right before OOM. i noticed that this dm-snapshot-ex is extremely high. slabinfo - version: 1.1 (SMP) kmem_cache 87 87 132 3 3 1 : 252 126 target_cmnd 0 0 60 0 0 1 : 252 126 iscsi_cmnd 0 0 124 0 0 1 : 252 126 dm-snapshot-in 128 180 40 2 2 1 : 252 126 dm-snapshot-ex 43236805 43237000 16 216185 216185 1 : 252 126 kcopyd-jobs 512 520 192 26 26 1 : 252 126 dm io 3328 3444 28 28 28 1 : 252 126 tcp_tw_bucket 0 0 128 0 0 1 : 252 126 tcp_bind_bucket 3 112 32 1 1 1 : 252 126 tcp_open_request 0 0 64 0 0 1 : 252 126 so under what kind of scenario, this dm-snapshot-ex can reach this high? i know the exception is used when some write on snapshot in 2.6.x, is this still hold true for 2.4.x? also i browse the patch and could not find where the dm-snapshot-ex is freed except when freeing the whole table. is this correct? thanks! Ming -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
