The high idle load is due to the use of polling instead of interrupts for card operations to ensure low latency. As data flow increases, this load shifts more towards useful work.
We recommend monitoring CPU load with the mpstat -P ALL utility.
grep 'expiration_date=' /etc/dpi/fastdpi.lic expiration_date=20991231
Format: YYYYMMDD
/etc/dpi/fastdpi.lic /etc/dpi/fastdpi.sig /etc/pf_ring/*
First option: Go to the required version from the main Wiki page, where announcements of new versions are published. Each version has a list of changes.
Second option: Install the yum-plugin-changelog package and use the changelog command:
yum install yum-plugin-changelog yum changelog 4 fastdpi
There is no version for FreeBSD. Only VEOS is supported.
We strongly recommend using the OS image specified in this article.
For monitoring, you can use solutions that utilize SNMP. For example, Zabbix Agent. Description
Cores are functionally distributed among various DPI tasks so they do not interfere with each other.
You can view the function distribution with the command:
ps -p pidof fastdpi H -o %cpu,lwp,psr,comm
DPI has service functions such as Netflow generation, Clickstream, PCAP recording, command processing, etc.
Their load is uneven and they can temporarily load a core to 100%, so a separate core is allocated to them to avoid interfering with traffic.
It may be due to the kipmi process of the remote server control interface ipmi, or perhaps the FW was not closed from external attacks.
When a process loads any core to 100%, softRAID stops working, making it impossible to access the server. There is also an article on other possible issues with ipmi: Kipmi0 eating up to 99.8% CPU on CentOS 6.4
The source code of the utilities is not available and is not planned to be provided. FreeBSD allows running native Linux applications. Also, for FreeBSD 9.2, a binary version of the utilities is available.