We have here a similar problem. A Linux box with Kernel 2.4.34 and
Openswan 2.4.6 have a redundantly internet connection. From some net's
we get all packets two times. That is not the normal condition but if
we transmit data through a tunnel to this net's we loses memory. The
skbuff_head_cache in /proc/slabinfo increase and the free memory decrease.
Active / Total Objects (% used) : 1244665 / 1247679 (99.8%)
Active / Total Slabs (% used) : 102622 / 102723 (99.9%)
Active / Total Caches (% used) : 42 / 58 (72.4%)
Active / Total Size (% used) : 394040.96K / 394690.33K (99.8%)
Minimum / Average / Maximum Object : 0.02K / 0.32K / 128.00K
OBJS ACTIVE USE OBJ SIZE SLABS OBJ/SLAB CACHE SIZE NAME
570980 570980 100% 0.19K 28549 21 114196K skbuff_head_cache
457950 457950 100% 0.25K 30530 16 122120K size-256
67670 67670 100% 2.00K 33835 2 135340K size-2048
45536 45536 100% 0.50K 5692 8 22768K size-512
total used free shared buffers cached
Mem: 515088 490672 24416 0 3276 46092
-/+ buffers/cache:441304 73784
If we stop the traffic skbuff_head_cache no longer increase. If the
increase the traffic the used skbuff_head_cache objects increase in the
same level.
If we prevent the double packages (i.e. disconnect on of the carrier
router) is stops increasing too.
Any one a hint what we can do?
regards
ralph
-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
------------------------------------------------------------------------
leaf-user mailing list: leaf-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/leaf-user
Support Request -- http://leaf-project.org/
