We have a Cisco 3845 router configured as a voice gateway with multi SIP trunks. But when it reachs 200 calls traffic, the CPU increase to 60-70% and caused by CCSIP_SPI_CONTROL process.
CPU utilization for five seconds: 46%/30%; one minute: 54%; five minutes: 58%
PID Runtime(ms) Invoked uSecs 5Sec 1Min 5Min TTY Process
377 400729448 171017979 2343 6.31% 10.71% 12.44% 0 CCSIP_SPI_CONTRO
365 89239288 197275072 452 2.55% 2.61% 2.72% 0 VOIP_RTCP
147 93017972 269045929 345 2.55% 2.51% 2.58% 0 IP Input
65 107469652 8634364 12446 1.19% 1.16% 1.17% 0 Per-Second Jobs
379 78432948 156592321 500 1.11% 1.94% 2.20% 0 CCSIP_UDP_SOCKET
According to the document from CISCO, this issue should be originated from DTMF is not RFC2833 and command "sip-kpml" in dial-peer instead of
dtmf-relay rtp-nte. But in our configuration, we use DTMF with RFC2833 for sure and dtmf-relay rtp-nte for all dial-peers and I don't think it's the main reason making high CPU.
Can you give me any suggesstion for this case? The IOS is : c3845-adventerprisek9-mz.151-3.T3.bin
Also, this is proccess memmory in our gw :
sh proc mem sor
Processor Pool Total: 343584124 Used: 151203668 Free: 192380456
I/O Pool Total: 40893952 Used: 14894528 Free: 25999424
PID TTY Allocated Freed Holding Getbufs Retbufs Process
0 0 103592788 45980208 44616904 0 0 *Init*
0 0 4216211956 2975264304 38130476 2622976 31460 *Dead*
1 0 2639240936 2625300112 26660328 0 0 Chunk Manager
377 0 1694209216 953055252 24561348 0 0 CCSIP_SPI_CONTRO
363 0 1654981196 1570893068 11654856 6804 0 CC-API_VCM
0 0 0 0 6472468 0 0 *MallocLite*
