Hi,
when using unionfs (20051130) for quite some time, the caches grow and
grow... and do not decrease. Slabtop:
Active / Total Objects (% used) : 1488446 / 1493583 (99.7%)
Active / Total Slabs (% used) : 40005 / 40005 (100.0%)
Active / Total Caches (% used) : 80 / 112 (71.4%)
Active / Total Size (% used) : 152251.51K / 152775.77K (99.7%)
Minimum / Average / Maximum Object : 0.01K / 0.10K / 128.00K
OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
477785 477546 99% 0.03K 4015 119 16060K size-32
452864 452527 99% 0.13K 15616 29 62464K dentry_cache
389840 389838 99% 0.04K 4430 88 17720K unionfs_dentry
84500 84496 99% 0.36K 8450 10 33800K unionfs_inode_cache
55830 55827 99% 0.39K 5583 10 22332K reiser_inode_cache
6188 5673 91% 0.27K 442 14 1768K radix_tree_node
4374 4075 93% 0.05K 54 81 216K buffer_head
3510 3135 89% 0.09K 78 45 312K vm_area_struct
2496 2402 96% 0.04K 26 96 104K sysfs_dir_cache
2263 2143 94% 0.12K 73 31 292K size-128
22:18 shanghai:~ > free
total used free shared buffers cached
Mem: 256784 252776 4008 0 16076 20428
-/+ buffers/cache: 216272 40512
Swap: 1036144 63908 972236
Note the unusually high 216272 - this should not have happened - someone
forgot to drop inodes (dput/iput?), because in normal operation, the
buffer/inode/etc. caches only grow as large as there is free RAM, and get
purged immediately if something new is put into RAM, like a malloc() from
userspace. Obviously, this is not the case (swap: 63908), since, as said,
someone forgot to release something so that pruning can't take place.
Upon unmounting the union mount, I get "VFS: Busy inodes. Self destruct in
5 seconds...", and even after that, it is _impossible_ to unmount the
underlying filesystem, even if all references are cleared.
Jan Engelhardt
--
| Alphagate Systems, http://alphagate.hopto.org/
| jengelh's site, http://jengelh.hopto.org/
_______________________________________________
unionfs mailing list
[email protected]
http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs
