upstream status not correct

[FrancisAston]

nginx version: openresty/1.19.9.1
nginx-module-vts-0.2.1/

server alread down，but the status page still up

nginx_upstream_check_module page

nginx-module-vts status page

[u5surf]

@FrancisAston Hi, Thanks reporting!!
Can you try once to change display format json? Because the State is changed by upstream_server_node.down, it changes in below code, but it might not to call when format is html.

nginx-module-vts/src/ngx_http_vhost_traffic_status_display_json.c

Lines 603 to 616 in 5b3812e

	#if (NGX_HTTP_UPSTREAM_CHECK)
	if (ngx_http_upstream_check_peer_down(peer->check_index)) {
	usn.down = 1;
	} else {
	usn.down = 0;
	}
	#else
	usn.down = (peer->fails >= peer->max_fails || peer->down);
	#endif
	#if nginx_version > 1007001
	usn.name = peer->name;
	#endif

I still don’t make sense obviously whether it is bug or not but it worth for diving deeply for this. I’m willing to investigate that.

[FrancisAston]

the 127.0.0.1:8540 is down , response json data: "down": false
I'm not a developer,Please help. thank a lot.
"ali-oss": [ {
"server": "127.0.0.1:8540",
"requestCounter": 0,
"inBytes": 0,
"outBytes": 0,
"responses": {
"1xx": 0,
"2xx": 0,
"3xx": 0,
"4xx": 0,
"5xx": 0
},
"requestMsecCounter": 0,
"requestMsec": 0,
"requestMsecs": {
"times": [],
"msecs": []
},
"requestBuckets": {
"msecs": [],
"counters": []
},
"responseMsecCounter": 0,
"responseMsec": 0,
"responseMsecs": {
"times": [],
"msecs": []
},
"responseBuckets": {
"msecs": [],
"counters": []
},
"weight": 1,
"maxFails": 3,
"failTimeout": 5,
"backup": false,
"down": false,
"overCounts": {
"maxIntegerSize": 18446744073709551615,
"requestCounter": 0,
"inBytes": 0,
"outBytes": 0,
"1xx": 0,
"2xx": 0,
"3xx": 0,
"4xx": 0,
"5xx": 0,
"requestMsecCounter": 0,
"responseMsecCounter": 0
}
}],

[u5surf]

@FrancisAston
Is this the below module which you use vts module together?
https://github.com/yaoweibin/nginx_upstream_check_module
And we appreciate if you can give us the nginx.conf which can reproduce this behavior.

[u5surf]

@FrancisAston
I've investigated, I seem to found the solution of this issue.
Can you check your nginx.conf whether upstream zone directive is active or not.
e.g.

http {
...
    vhost_traffic_status_zone;
    upstream web {
            zone backend 64k; # <- here
            check interval=5000 rise=1 fall=3 timeout=4000;
            server 127.0.0.1:9000;
    }
    server {
        listen       8000;
        location / {
            proxy_pass http://web;
        }
    }
    server {
        listen       80;
        location /status/ {
            vhost_traffic_status_display;
            vhost_traffic_status_display_format json; #prometheus;
        }
    }
}

c.f. http://nginx.org/en/docs/http/ngx_http_upstream_module.html#zone

if the directive is inactive, the upstream alive status check which I referred the comment in #258 (comment) can't pass for the goto is exist before.

nginx-module-vts/src/ngx_http_vhost_traffic_status_display_json.c

Lines 573 to 575 in 5b3812e

	if (uscf->shm_zone == NULL) {
	goto not_supported;
	}

As a test we activate the zone directive, the status down can return true in case of upstream is dead.

result

% curl -s http://localhost/status/ | jq '.upstreamZones.web[0].down'
true

gdb info:

Breakpoint 1, ngx_http_vhost_traffic_status_display_set_upstream_group (r=0x558bb9bb7180, buf=0x558bb9c094fe "") at ../ng
inx-module-vts/src/ngx_http_vhost_traffic_status_display_json.c:573
573                 if (uscf->shm_zone == NULL) {
(gdb) p uscf->shm_zone
$1 = (ngx_shm_zone_t *) 0x558bb9bc4320
(gdb) c

[FrancisAston]

http {
...
vhost_traffic_status_zone;
upstream web {
zone backend 64k; # <- here
check interval=5000 rise=1 fall=3 timeout=4000;
server 127.0.0.1:9000;
}

add zone backend 64k; and reload nginx.
it work. thinks 。