惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

美团技术团队
T
The Blog of Author Tim Ferriss
月光博客
月光博客
阮一峰的网络日志
阮一峰的网络日志
Engineering at Meta
Engineering at Meta
量子位
I
InfoQ
Jina AI
Jina AI
Microsoft Security Blog
Microsoft Security Blog
H
Help Net Security
H
Hackread – Cybersecurity News, Data Breaches, AI and More
G
Google Developers Blog
J
Java Code Geeks
Recent Announcements
Recent Announcements
aimingoo的专栏
aimingoo的专栏
小众软件
小众软件
V
V2EX
腾讯CDC
P
Proofpoint News Feed
A
About on SuperTechFans
爱范儿
爱范儿
U
Unit 42
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
Last Week in AI
Last Week in AI

CoreDNS: DNS and Service Discovery

CoreDNS-1.14.2 Release CoreDNS-1.14.1 Release CoreDNS-1.14.0 Release CoreDNS-1.13.2 Release CoreDNS-1.13.1 Release CoreDNS-1.13.0 Release CoreDNS-1.12.4 Release CoreDNS-1.12.3 Release CoreDNS-1.12.2 Release CoreDNS-1.12.1 Release CoreDNS-1.12.0 Release CoreDNS-1.11.4 Release CoreDNS-1.11.3 Release k8s_cache CoreDNS-1.11.2 Release CoreDNS-1.11.1 Release CoreDNS-1.11.0 Release CoreDNS-1.10.1 Release CoreDNS-1.10.0 Release CoreDNS-1.9.4 Release CoreDNS-1.9.3 Release CoreDNS-1.9.2 Release CoreDNS-1.9.1 Release CoreDNS and Apache APISIX open new doors for Service Discovery? Trail Of Bits Security Review CoreDNS-1.9.0 Release CoreDNS-1.8.7 Release CoreDNS-1.8.6 Release CoreDNS-1.8.5 Release CoreDNS-1.8.4 Release
ebpf
2021-10-14 · via CoreDNS: DNS and Service Discovery

Description

This experimental plugin allows you to use an eBPF XDP program to analyze and filter traffic before it reaches CoreDNS, and report very basic Prometheus metrics. When CoreDNS exits, the program will be detached from the interface.

This generic solution serves in part as an example of how you can integrate an eBPF XDP program with CoreDNS with a custom plugin. But due to the generic nature, map entry is somewhat cryptic and metrics must be defined in the Corefile, limiting their scope. When writing your own plugin, you can tailor it to work with a specific XDP program, for example, to enable easier human-readable data entry or publish more advanced metrics.

Syntax

ebpf {
  elf PROGRAM
  if INTERFACE
  map [KEY] VALUE
  metric NAME KEY POS LEN "HELP"
}
  • elf PROGRAM - the ELF program to attach. See notes below on program requirements.
  • if INTERFACE - the interface to attach to
  • map KEY VALUE - the hexidecimal string representations of the KEY and VALUE of an entry to load into the eBPF map. You may specify the map option more than once to add multiple items to the map. If KEY is not specified, the entry is treated as an array value. To make multi-field values easier to visually digest, VALUE may be delimited by dots. e.g. 012345678.0000000000000000.9ABCDEF0 This is for legibility of the Corefile only; any dots in VALUE are ignored by the parser. When debug is used the values written to log are not delimited.
  • metric NAME KEY POS LENHELP” - when used in conjunction with the prometheus plugin, register a Prometheus “gauge” metric to expose a eBPF map value as an integer metric. The metric is named NAME with help text of HELP. The map value to use is determined by the KEY, byte position POS, and length LEN in bytes. LEN can be at most 8 bytes (64 bits). The integer value should be little endian.

Please be aware of the considerable footgun potential of this plugin. An XDP program attached to an interface will act on all ingress packets to the interface - not just packets bound for CoreDNS.

eBPF Program and Map Requirements

The program must be an XDP program, and main function named xdp_prog. The map must be named xdp_map.

Some example programs written in C are included in https://github.com/InfobloxOpen/ebpf/tree/master/example_programs.

Examples

If my_xdp_program.o defines a map with a 4 byte key, and the following struct as a value …

struct maprec {
  __be32  ip4net;  // ipv4 network
  __be32  ip4mask; // ipv4 mask
  __be32  count;   // packet count
};

The following will attach my_xdp_program.o to eth0, and load data for IP network 10.11.0.0 , IP mask 255.255.0.0, and a count of zero (0A0B0000, FFFF0000, and 00000000 respectively) into key 00000000 of the map.

. {
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 00000000 0A0B0000FFFF000000000000
  }
}

The following adds dots to the map value to make it easier to read.

. {
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 00000000 0A0B0000.FFFF0000.00000000
  }
}

The following will enable debug to monitor map values and log when they change.

. {
  debug
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 00000000 0A0B0000.FFFF0000.00000000
  }
}

The following adds map entries without specifying keys. Each map entry is inserted as an array value, with an automatically incrementing key.

. {
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 0A0B0000.FFFF0000.00000000
    map 0A0C0000.FFFF0000.00000000
    map 0A0D0000.FFFF0000.00000000
  }
}

The example above is equivalent to the following but with keys specified. Note that the keys are little endian in this example.

. {
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 00000000 0A0B0000.FFFF0000.00000000
    map 01000000 0A0C0000.FFFF0000.00000000
    map 02000000 0A0D0000.FFFF0000.00000000
  }
}

The following exposes a Prometheus metric. The metric is named coredns_ebpf_example_total and the value will reflect the rightmost 4 bytes from map entry 02000000.

. {
  prometheus :9153
  ebpf {
    if eth0
    elf my_xdp_program.o
    map 00000000 0A0B0000.FFFF0000.00000000
    map 01000000 0A0C0000.FFFF0000.00000000
    map 02000000 0A0D0000.FFFF0000.00000000
    metric example_total 02000000 8 4 "Example count."
  }
}