请问metrics里的err统计规则是什么？ #424
-
|
版本：4.1.4 运行了一天，发现metrics里err非常多，请问这是为什么？配置在最底下 {
"plugins": {
"_default_cache": {
"hit": 0,
"lazy_hit": 0,
"query": 0,
"size": 0
},
"cache": {
"hit": 325,
"lazy_hit": 0,
"query": 4889,
"size": 58
},
"forward_local": {
"0": {
"err": 1921,
"latency": {
"avg": 19,
"max": 631,
"min": 0,
"p25": 4,
"p50": 6,
"p75": 8.75
},
"query": 2169
},
"1": {
"err": 1414,
"latency": {
"avg": 98,
"max": 3374,
"min": 0,
"p25": 19,
"p50": 26,
"p75": 44
},
"query": 2169
}
},
"forward_remote": {
"0": {
"err": 1158,
"latency": {
"avg": 232,
"max": 2489,
"min": 46,
"p25": 75,
"p50": 152,
"p75": 263.75
},
"query": 2213
},
"1": {
"err": 2207,
"latency": {
"avg": 276,
"max": 509,
"min": 126,
"p25": 130,
"p50": 168,
"p75": 476.5
},
"query": 2213
}
}
},
"servers": {}
}log:
level: error
file: "dns.log"
data_providers:
- tag: geosite
file: ./geosite.dat
auto_reload: true
- tag: geoip
file: ./geoip-only-cn-private.dat
auto_reload: true
plugins:
# 缓存
- tag: cache
type: cache
args:
size: 1024
# 转发至本地服务器的插件
- tag: forward_local
type: fast_forward
args:
upstream:
- addr: 'tls://dot.pub'
dial_addr: '120.53.53.53:853'
enable_pipeline: false
max_conns: 2
trusted: true
- addr: 'tls://b.iqiq.io'
dial_addr: '101.42.236.152:853'
enable_pipeline: false
max_conns: 2
trusted: true
# 转发至远程服务器的插件
- tag: forward_remote
type: fast_forward
args:
upstream:
- addr: 'https://public.dns.iij.jp/dns-query'
dial_addr: '103.2.57.5:443'
max_conns: 2
trusted: true
- addr: 'https://dns-unfiltered.adguard.com/dns-query'
dial_addr: '94.140.14.141:443'
max_conns: 2
trusted: true
# 匹配本地域名的插件
- tag: query_is_local_domain
type: query_matcher
args:
domain:
- 'provider:geosite:cn'
# 匹配非本地域名的插件
- tag: query_is_non_local_domain
type: query_matcher
args:
domain:
- 'provider:geosite:geolocation-!cn'
# 匹配广告域名的插件
- tag: query_is_ad_domain
type: query_matcher
args:
domain:
- 'provider:geosite:category-ads-all'
# 匹配本地 IP 的插件
- tag: response_has_local_ip
type: response_matcher
args:
ip:
- 'provider:geoip:cn'
- tag: main_sequence
type: sequence
args:
exec:
- _misc_optm
- cache
- if: query_is_ad_domain
exec:
- _new_nxdomain_response
- _return
- if: query_is_local_domain
exec:
- forward_local
- _return
- if: query_is_non_local_domain
exec:
- _prefer_ipv4
- forward_remote
- _return
-
- primary:
- forward_local
- if: "(! response_has_local_ip) && [_response_valid_answer]"
exec:
- _drop_response
secondary:
- _prefer_ipv4
- forward_remote
fast_fallback: 1000
servers:
- exec: main_sequence
listeners:
- protocol: udp
addr: 127.0.0.1:53
- protocol: udp
addr: '[::1]:53'
api:
http: "127.0.0.1:58080" |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 4 replies
-
|
开 warn log 看。 |
Beta Was this translation helpful? Give feedback.
-
|
打开warn log之后，DoH的报出了这个错误，推测应该是一段时间没有dns查询，被远程服务器关闭了链接吧。 DoT也有err，但是没有日志。不过应该也是因为这个原因 {
"plugins": {
"_default_cache": {
"hit": 0,
"lazy_hit": 0,
"query": 0,
"size": 0
},
"cache": {
"hit": 18,
"lazy_hit": 0,
"query": 165,
"size": 14
},
"forward_local": {
"0": {
"err": 23,
"latency": {
"avg": 13,
"max": 42,
"min": 5,
"p25": 7,
"p50": 7,
"p75": 14.5
},
"query": 37
},
"1": {
"err": 14,
"latency": {
"avg": 23,
"max": 66,
"min": 4,
"p25": 13,
"p50": 23,
"p75": 26
},
"query": 37
}
},
"forward_remote": {
"0": {
"err": 4,
"latency": {
"avg": 151,
"max": 308,
"min": 47,
"p25": 72.75,
"p50": 138,
"p75": 230
},
"query": 90
}
}
},
"servers": {}
} |
Beta Was this translation helpful? Give feedback.
-
|
正常情况 err 率不会高。 参考 dnspod 的 tls "forward_local": {
"0": {
"err": 8,
"latency": {
"avg": 174,
"max": 1186,
"min": 36,
"p25": 103.25,
"p50": 138,
"p75": 180
},
"query": 85913
}
},err 的统计方式很简单: 请求发出去了，但没有收到应答。 err 率高多半是网络问题。 |
Beta Was this translation helpful? Give feedback.
-
|
好的，感谢回复。 我再开debug日志看下是啥原因 |
Beta Was this translation helpful? Give feedback.
-
|
好像发现了一点端倪，我在配置文件里 forward_local只配置一个上游(tls://dns.alidns.com)的时候，本地执行nslookup发送几个dns解析请求，统计里的err就是0。 但是配置两个上游时，日志就变成了这样，其中一个上游的query和err一样。 {
"plugins": {
"_default_cache": {
"hit": 0,
"lazy_hit": 0,
"query": 0,
"size": 0
},
"cache": {
"hit": 0,
"lazy_hit": 0,
"query": 10,
"size": 4
},
"forward_local": {
"0": {
"err": 0,
"latency": {
"avg": 11,
"max": 40,
"min": 6,
"p25": 6,
"p50": 6,
"p75": 11
},
"query": 10
},
"1": {
"err": 10,
"latency": {
"avg": 0,
"max": 0,
"min": 0,
"p25": 0,
"p50": 0,
"p75": 0
},
"query": 10
}
},
"forward_remote": {
"0": {
"err": 0,
"latency": {
"avg": 174,
"max": 229,
"min": 119,
"p25": 119,
"p50": 174,
"p75": 229
},
"query": 2
}
}
},
"servers": {}
} |
Beta Was this translation helpful? Give feedback.
-
|
确实。同一插件配置多上游的时候计数有bug。 |
Beta Was this translation helpful? Give feedback.
正常情况 err 率不会高。
参考 dnspod 的 tls
err 的统计方式很简单: 请求发出去了，但没有收到应答。
err 率高多半是网络问题。