i all
I use linux as GiGE router and have 6 NIC on it
Those days the NIC interrupt takes around 100% CPU but the system is 4G memroy
and 8 CPU. I can't see any error packet in this NIC interface too
After I block the udp, the %CPU drops. but the UDP only takes around 8M in
general
We use UDP traffic for voice.
Do you have any suggestion ? increase the kernel parameter?
Thank you so much
---------------------------------
Be smarter than spam. See how smart SpamGuard is at giving junk email the boot
with the All-new Yahoo! Mail
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.centos.org/pipermail/centos/attachments/20071011/064bfde1/attachment-0004.html>
> I use linux as GiGE router and have 6 NIC on it > Those days the NIC interrupt takes around 100% CPU but the system is 4G memroy and 8 CPU. I can't see any error packet in this NIC interface too > > After I block the udp, the %CPU drops. but the UDP only takes around 8M in general > > We use UDP traffic for voice. > > Do you have any suggestion ? increase the kernel parameter?Add the following lines to /etc/sysctl.conf net.core.rmem_default = 67108864 net.core.wmem_default = 67108864 net.core.rmem_max = 67108864 net.core.wmem_max = 67108864 net.ipv4.tcp_mem = 4096 67108864 67108864 net.ipv4.tcp_rmem = 4096 67108864 67108864 net.ipv4.tcp_wmem = 4096 67108864 67108864 net.ipv4.ip_local_port_range = 32768 65535 net.ipv4.tcp_max_syn_backlog = 8192 After adding these lines, run "sysctl -p"
-----Original Message----- From: centos-bounces at centos.org [mailto:centos-bounces at centos.org] On Behalf Of chloe K i all I use linux as GiGE router and have 6 NIC on it Those days the NIC interrupt takes around 100% CPU but the system is 4G memroy and 8 CPU. I can't see any error packet in this NIC interface too After I block the udp, the %CPU drops. but the UDP only takes around 8M in general Is there any chance you have a routing loop problem? when your CPU jumps to 100% usage, is traffic still passing, and your applications working? (I am also assuming that all 8CPU's are at 100% usage) are you blocking the UDP traffic with iptables on your linux box? if so...maybe set it to be a log rule instead, then analyze those logs. as a side note, when your interrupts are taking 100% CPU...it is called "Live Lock" (in case you didn't know) We use UDP traffic for voice. Do you have any suggestion ? increase the kernel parameter? I take it that there was no interesting information in 'dmesg' as well. Mike -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.centos.org/pipermail/centos/attachments/20071011/dd865a04/attachment-0004.html>