[Swan] Pluto consumes all available memory

Will Roberts ironwill42 at gmail.com
Sun May 31 17:33:13 EEST 2015


On 05/20/2015 04:00 PM, Paul Wouters wrote:
> Possible yes, but that would be very strange. Do you see memory usage go
> up over time gradually, or in giant chunks?

The system loses about 6MB a day. Many of my servers are relatively 
small with about 512MB of RAM. The larger machines are also seeing the 
problem, they just take longer to hit the critical point.

>
> Nothing I can think of. You can try to run things under valgrind I guess
> and see?
>
> ipsec _stackmanager start
> valgrind /usr/libexec/ipsec/pluto --config /etc/ipsec.conf --nofork
>
> Pssobly when you hit ctrl-c valgrind will give some data?

Here's what valgrind has to say. I'm going to re-run with the full leak 
check and -v

==3687== HEAP SUMMARY:
==3687==     in use at exit: 56,104,711 bytes in 61,144 blocks
==3687==   total heap usage: 8,282,886 allocs, 8,221,742 frees, 
7,522,692,000 bytes allocated
==3687==
==3687== LEAK SUMMARY:
==3687==    definitely lost: 1,097,141 bytes in 8,575 blocks
==3687==    indirectly lost: 530,470 bytes in 60 blocks
==3687==      possibly lost: 32,999 bytes in 117 blocks
==3687==    still reachable: 54,444,101 bytes in 52,392 blocks
==3687==         suppressed: 0 bytes in 0 blocks
==3687== Rerun with --leak-check=full to see details of leaked memory
==3687==
==3687== For counts of detected and suppressed errors, rerun with: -v
==3687== Use --track-origins=yes to see where uninitialised values come from
==3687== ERROR SUMMARY: 17167 errors from 2 contexts (suppressed: 28 from 9)

--Will


More information about the Swan mailing list