After a week without problem today pol crash two times. After long time of normal memory usage happen that in few second memory raise until 2 gb and then hang on for about 25 minutes (in this time no connections to server is possible). After that pol shutdown. At riup everything appears normal.
pol.log report something like this for a few time. Is always a different last script but i noticed that is always something about hook packets(the most made by me with packet guide).
Same hooks was running without problem for month under older version of pol96(08-12-05).
[08/28 16:00:21] sysload=100 (20218) cputime=0
*Thread Info*
Semaphore PID: 2596
("kill -SIGUSR2 2596" to output backtrace)
Scripts Thread Checkpoint: 112
Last Script: pkg/packet/hooks_gumps/sendskill.ecl PC: 19
Escript Instruction Cycles: -58333541
Tasks Thread Checkpoint: 1
Active Client Thread Checkpoint: 116
Current Threads:
2555 - Main
2593 - Listen
2594 - HTTP
2595 - Tasks
2596 - Scripts
2598 - Decay_malas
2599 - Decay_tokuno
2601 - Decay_ilshenar
2602 - Decay_britannia
2603 - Reap
2604 - DbgListn
2605 - ThreadStatus
2606 - Console
27730 - Client#1513
28114 - Client#1584
28343 - Client#1640
28590 - Client#1712
28609 - Client#1729
28628 - Client#1734
28755 - Client#1790
28763 - Client#1794
28778 - Client#1796
28847 - Client#1810
28853 - Client#1814
28854 - Client#1815
28855 - Client#1816
28860 - Client#1817
28862 - Client#1818
28865 - Client#1820
28868 - Client#1821
28870 - Client#1823
28874 - Client#1826
28875 - Client#1827
28877 - Client#1828
28897 - Client#1829
28904 - http_conn_thread:17
28908 - http_conn_thread:26
28912 - http_conn_thread:35
28913 - http_conn_thread:38
28915 - http_conn_thread:40
28916 - http_conn_thread:41
28917 - http_conn_thread:42
28918 - http_conn_thread:43
Everything appears normal in script.log
No dump generated. Any suggestion? I'm an idiot(it's really possibile

)?
Server: Linux