Ben0625 opened a new issue #4103:
URL: https://github.com/apache/apisix/issues/4103
### Issue description
One request with unhealthy code returned will cause multiple unhealthy HTTP
increment if the size of passive.unhealthy.http_statuses is more than 1.
### Environment
* apisix version: 2.0
* OS: 3.10.0-693.el7.x86_64
* OpenResty version: openresty/1.17.8.2
* etcd version: 3.4.13
* apisix-dashboard version: 2.0
### Minimal test code / Steps to reproduce the issue
The route configuration in etcd is below:
```
{
"id": "route_2050210843",
"create_time": 1618455175,
"update_time": 1619013202,
"uris": ["/test1.do"],
"name": "route-2050210843",
"desc": "health_check_test",
"methods": ["GET", "POST", "PUT", "DELETE", "PATCH", "HEAD", "OPTIONS"],
"vars": [],
"upstream": {
"nodes": [{
"host": "172.25.69.182",
"port": 8080,
"weight": 1
}, {
"host": "172.25.70.116",
"port": 8080,
"weight": 1
}],
"timeout": {
"connect": 6,
"read": 6,
"send": 6
},
"type": "roundrobin",
"checks": {
"active": {
"healthy": {
"interval": 100,
"successes": 10
},
"host": "172.25.69.182",
"http_path": "/check.do",
"req_headers": [""],
"timeout": 3,
"unhealthy": {
"http_failures": 5,
"interval": 3
}
},
"passive": {
"healthy": {
"http_statuses": [200],
"successes": 2
},
"unhealthy": {
"http_failures": 5,
"http_statuses": [500, 502, 504],
"tcp_failures": 1
}
}
}
},
"labels": {
"serve": "2050210843"
}
}
```
1. Set a route with two upstreams and healthcheck active and passive
configuration. One upstream always return 500 and another always return 200.
2. Send a request to trigger creating new checker. It will be passed to the
normal upstream.
3. Send a second request, it will be passed to the unnormal upstream.
### What's the actual result?
One request with unnormal upstream will cause three times unhealthy HTTP
increment.
`2021/04/21 21:54:00 [warn] 73450#73450: *3017774 [lua]
healthcheck.lua:1142: log(): [healthcheck]
(upstream#/apisix/routes/route_2050210843) unhealthy HTTP increment (1/5) for
'172.25.69.182(172.25.69.182:8080)' while logging request, client:
10.58.50.104, server: , request: "GET /test1.do HTTP/1.1", upstream:
"http://172.25.69.182:8080/test1.do";, host: "172.25.71.28:9080"`
`2021/04/21 21:54:00 [warn] 73450#73450: *3017774 [lua]
healthcheck.lua:1142: log(): [healthcheck]
(upstream#/apisix/routes/route_2050210843) unhealthy HTTP increment (2/5) for
'172.25.69.182(172.25.69.182:8080)' while logging request, client:
10.58.50.104, server: , request: "GET /test1.do HTTP/1.1", upstream:
"http://172.25.69.182:8080/test1.do";, host: "172.25.71.28:9080"`
`2021/04/21 21:54:00 [warn] 73450#73450: *3017774 [lua]
healthcheck.lua:1142: log(): [healthcheck]
(upstream#/apisix/routes/route_2050210843) unhealthy HTTP increment (3/5) for
'172.25.69.182(172.25.69.182:8080)' while logging request, client:
10.58.50.104, server: , request: "GET /test1.do HTTP/1.1", upstream:
"http://172.25.69.182:8080/test1.do";, host: "172.25.71.28:9080"`
### What's the expected result?
One request with unnormal upstream will cause only once unhealthy HTTP
increment.
I cannot understand why it has to be double loop in function
`healcheck_passive` of apisix/init.lua
```
for i, status in ipairs(http_statuses) do
for i, status in ipairs(http_statuses) do
if resp_status == status then
checker:report_http_status(api_ctx.balancer_ip,
port or api_ctx.balancer_port,
host,
resp_status)
end
end
end
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
