Issue description
The APISIX prometheus plugin uses the /nginx_status endpoint provided by nginx to get some internal metrics, which return the data like
Active connections: 291
server accepts handled requests
16630948 16630948 31070465
Reading: 6 Writing: 179 Waiting: 106
According to the nginx documentation:
The following status information is provided:
Active connections
The current number of active client connections including Waiting connections.
accepts
The total number of accepted client connections.
handled
The total number of handled connections. Generally, the parameter value is the same as accepts unless some resource limits have been reached (for example, the worker_connections limit).
requests
The total number of client requests.
Reading
The current number of connections where nginx is reading the request header.
Writing
The current number of connections where nginx is writing the response back to the client.
Waiting
The current number of idle client connections waiting for a request.
The third number in the third line is The total number of client requests, but APISIX now puts this metric in nginx_http_current_connections{label="total"}, which is obviously wrong.
|
metrics.connections = prometheus:gauge("nginx_http_current_connections", |
|
"Number of HTTP connections", |
|
{"state"}) |
|
local ngx_status_items = {"active", "accepted", "handled", "total", |
|
"reading", "writing", "waiting"} |
|
local label_values = {} |
|
|
|
local function nginx_status() |
|
local res = ngx_capture("/apisix/nginx_status") |
|
if not res or res.status ~= 200 then |
|
core.log.error("failed to fetch Nginx status") |
|
return |
|
end |
|
|
|
-- Active connections: 2 |
|
-- server accepts handled requests |
|
-- 26 26 84 |
|
-- Reading: 0 Writing: 1 Waiting: 1 |
|
|
|
local iterator, err = re_gmatch(res.body, [[(\d+)]], "jmo") |
|
if not iterator then |
|
core.log.error("failed to re.gmatch Nginx status: ", err) |
|
return |
|
end |
|
|
|
for _, name in ipairs(ngx_status_items) do |
|
local val = iterator() |
|
if not val then |
|
break |
|
end |
|
|
|
label_values[1] = name |
|
metrics.connections:set(val[0], label_values) |
|
|
|
end |
|
end |
Also the three metrics in the third line are all cumulative, should we use the counter type instead of the guage for those metrics? Which is more in line with the prometheus definition of metric types.
So I want to introduce two new metric:
nginx_http_requests_total(counter type)nginx_http_connections_total(counter type) with a type label, which has two values: accepts, handled
If there are no problems, I can complete the fix.
Environment
- apisix version (cmd:
apisix version): 2.10.1
Steps to reproduce
Just enable prometheus plugin.
Actual result
/
Error log
/
Expected result
No response
Issue description
The APISIX prometheus plugin uses the
/nginx_statusendpoint provided by nginx to get some internal metrics, which return the data likeAccording to the nginx documentation:
The third number in the third line is
The total number of client requests, but APISIX now puts this metric innginx_http_current_connections{label="total"}, which is obviously wrong.apisix/apisix/plugins/prometheus/exporter.lua
Lines 86 to 88 in 0920d1d
apisix/apisix/plugins/prometheus/exporter.lua
Lines 174 to 206 in 0920d1d
Also the three metrics in the third line are all cumulative, should we use the counter type instead of the guage for those metrics? Which is more in line with the prometheus definition of metric types.
So I want to introduce two new metric:
nginx_http_requests_total(counter type)nginx_http_connections_total(counter type) with atypelabel, which has two values:accepts,handledIf there are no problems, I can complete the fix.
Environment
apisix version): 2.10.1Steps to reproduce
Just enable prometheus plugin.
Actual result
/
Error log
/
Expected result
No response