Please add support for display format prometheus

[zhouchangxun]

Please add support for display format prometheus

[zhouchangxun]

@Frodox Hi, Thanks for your suggestion.
I'll get it checked when i'm free.

[taomaree]

alibaba/tengine#1426

[zhouchangxun]

@taomaree @Frodox
I added this feature (cdc74da) manually based on what you mentioned (alibaba/tengine#1426 )
By the way, the output format is different from the raw one ( I added a property upstream_type for every record) because the output contain both http upstream and stream upstream.
$ curl 'http://localhost:8081/status?format=prometheus&status=down'

# HELP nginx_upstream_count_total Nginx total number of servers
# TYPE nginx_upstream_count_total gauge
nginx_upstream_count_total 6
# HELP nginx_upstream_count_up Nginx total number of servers that are UP
# TYPE nginx_upstream_count_up gauge
nginx_upstream_count_up 0
# HELP nginx_upstream_count_down Nginx total number of servers that are DOWN
# TYPE nginx_upstream_count_down gauge
nginx_upstream_count_down 6
# HELP nginx_upstream_count_generation Nginx generation
# TYPE nginx_upstream_count_generation gauge
nginx_upstream_count_generation 1
# HELP nginx_upstream_server_rise Nginx rise counter
# TYPE nginx_upstream_server_rise counter
nginx_upstream_server_rise{index="0",upstream_type="http",upstream="http-cluster",name="127.0.0.1:8082",status="down",type="http",port="0"} 0
nginx_upstream_server_rise{index="1",upstream_type="http",upstream="http-cluster",name="127.0.0.2:8082",status="down",type="http",port="0"} 0
nginx_upstream_server_rise{index="1",upstream_type="stream",upstream="tcp-cluster",name="192.168.0.2:22",status="down",type="tcp",port="0"} 0
nginx_upstream_server_rise{index="2",upstream_type="stream",upstream="udp-cluster",name="127.0.0.1:5432",status="down",type="udp",port="0"} 0
nginx_upstream_server_rise{index="4",upstream_type="stream",upstream="http-cluster2",name="127.0.0.1:8082",status="down",type="http",port="0"} 0
nginx_upstream_server_rise{index="5",upstream_type="stream",upstream="http-cluster2",name="127.0.0.2:8082",status="down",type="http",port="0"} 0
# HELP nginx_upstream_server_fall Nginx fall counter
# TYPE nginx_upstream_server_fall counter
nginx_upstream_server_fall{index="0",upstream_type="http",upstream="http-cluster",name="127.0.0.1:8082",status="down",type="http",port="0"} 41
nginx_upstream_server_fall{index="1",upstream_type="http",upstream="http-cluster",name="127.0.0.2:8082",status="down",type="http",port="0"} 42
nginx_upstream_server_fall{index="1",upstream_type="stream",upstream="tcp-cluster",name="192.168.0.2:22",status="down",type="tcp",port="0"} 14
nginx_upstream_server_fall{index="2",upstream_type="stream",upstream="udp-cluster",name="127.0.0.1:5432",status="down",type="udp",port="0"} 40
nginx_upstream_server_fall{index="4",upstream_type="stream",upstream="http-cluster2",name="127.0.0.1:8082",status="down",type="http",port="0"} 40
nginx_upstream_server_fall{index="5",upstream_type="stream",upstream="http-cluster2",name="127.0.0.2:8082",status="down",type="http",port="0"} 43
# HELP nginx_upstream_server_active Nginx active 1 for UP / 0 for DOWN
# TYPE nginx_upstream_server_active gauge
nginx_upstream_server_active{index="0",upstream_type="http",upstream="http-cluster",name="127.0.0.1:8082",type="http",port="0"} 0
nginx_upstream_server_active{index="1",upstream_type="http",upstream="http-cluster",name="127.0.0.2:8082",type="http",port="0"} 0
nginx_upstream_server_active{index="1",upstream_type="stream",upstream="tcp-cluster",name="192.168.0.2:22",type="tcp",port="0"} 0
nginx_upstream_server_active{index="2",upstream_type="stream",upstream="udp-cluster",name="127.0.0.1:5432",type="udp",port="0"} 0
nginx_upstream_server_active{index="4",upstream_type="stream",upstream="http-cluster2",name="127.0.0.1:8082",type="http",port="0"} 0
nginx_upstream_server_active{index="5",upstream_type="stream",upstream="http-cluster2",name="127.0.0.2:8082",type="http",port="0"} 0

Please check if it can meet your requirement (To be honest, I am not familiar with prometheus format :D )

[Frodox]

@zhouchangxun can we cherry-pick commits from alibaba/tengine#1426 ? :D

[Frodox]

@zhouchangxun looks good 👍
but..what is the port label for? why is it 0 every where ?

[hroost]

@Frodox it's 0 when port (at check directive) is not specially defined and module using port value from upstream server