help request: The route occasionally returns a 404 page not found

[wangfeng22]

Description

I have configured several routes, and everything going well. But when I added a new route, sometimes this route returned 404 page not found. That's really abnormal.In fact, the back-end didn't handle the 404 requests.
I have restarted apisix, but this problem still exists. Only the domain name added later has this problem.

ps: The Route support http and https. The Upstream is configured like http://domain

Environment

• APISIX version (run apisix version): 2.13.2
• Operating system (run uname -a):
• OpenResty / Nginx version (run openresty -V or nginx -V):
• etcd version, if relevant (run curl http://127.0.0.1:9090/v1/server_info):
• APISIX Dashboard version, if relevant:
• Plugin runner version, for issues related to plugin runners:
• LuaRocks version, for installation issues (run luarocks --version):

[monkeyDluffy6017]

Could you provide a minimal reproduction use case with detailed steps？

[tzssangglass]

maybe fix by: #7466 and release in v2.15.0

[wangfeng22]

maybe fix by: #7466 and release in v2.15.0

I have studied this issue for many days, and I find it difficult to understand the explanation of this problem. Why can't I restart Apisix to solve this problem. @tzssangglass

[wangfeng22]

Could you provide a minimal reproduction use case with detailed steps？

1. add several routes like routerA, routerB, routerC, and so on. The router support HTTP and HTTPS. The upstream is configured like http://domain.
2. add routerD. Then a strange thing happened, routerD occasionally returns 404 page not found. In fact, the upstream of routerD don't handle the 404 requests.

I don't know what returns 404 page not found. routerA, routerB and routerC are working well. I have restarted apisix, but this problem still exists. @monkeyDluffy6017

[wangfeng22]

maybe fix by: #7466 and release in v2.15.0

The 404 response is returned right now, and it doesn't get a response from upstream. You can see it from No number.
@tzssangglass

[wangfeng22]

@tokers Could you help me see see? Looking forward to your reply.

[tzssangglass]

The 404 response is returned right now, and it doesn't get a response from upstream. You can see it from No number.

Are you saying that this problem still exists in 2.15.0? Or are you insisting on reproducing and working on this issue on 2.13.2?

[wangfeng22]

I insist on reproducing and working on this issue on 2.13.2. I really want to know why this happened. This problem can be reproduced in the case of pressure measurement. It is difficult to understand the explanation of this problem in this issue #7466 @tzssangglass

[monkeyDluffy6017]

I checked the code, and will show you how this happened.
If without this fix #7466, we assume that the first request is https, and the second request is http. we will get pool_opt: {\"pool\":\"127.0.0.1#1983#a.com\",\"pool_size\":4} in req1, get pool_opt: {\"pool\":\"127.0.0.1#1983#a.com\",\"pool_size\":8} in req2, the http req will get the https req's pool, and it's not expected. the http req will use the connection from the https connection pool, and get 404 response.

[monkeyDluffy6017]

The scenario you listed is the most basic function of apisix, if there is a problem with this scenario, there will be a lot of user feedback, did you miss any condition?
Do you have keepalive enabled? If so, can you turn off keepalive and try it out?

            local code, body = test('/apisix/admin/upstreams/1',
                ngx.HTTP_PUT,
                [[{
                    "scheme": "https",
                    "type": "roundrobin",
                    "nodes": {
                        "127.0.0.1:1983": 1
                    },
                    "pass_host": "rewrite",
                    "upstream_host": "a.com",
                    "keepalive_pool": {                            <=========== here
                        "size": 4
                    }
                }]]
            )

[wangfeng22]

The scenario you listed is the most basic function of apisix, if there is a problem with this scenario, there will be a lot of user feedback, did you miss any condition? Do you have keepalive enabled? If so, can you turn off keepalive and try it out?

            local code, body = test('/apisix/admin/upstreams/1',
                ngx.HTTP_PUT,
                [[{
                    "scheme": "https",
                    "type": "roundrobin",
                    "nodes": {
                        "127.0.0.1:1983": 1
                    },
                    "pass_host": "rewrite",
                    "upstream_host": "a.com",
                    "keepalive_pool": {                            <=========== here
                        "size": 4
                    }
                }]]
            )

I have configured keepalive_timeout as 0 in config.yaml, but keepalive_pool is configured as defalut by apisix-dashboard.
@monkeyDluffy6017

[monkeyDluffy6017]

use curl instead of apisix-dashboard

[wangfeng22]

The following is upstream config in etcd.

{"hash_on":"vars","desc":"api-add","type":"roundrobin","keepalive_pool":{"size":320,"idle_timeout":60,"requests":1000},"timeout":{"read":600,"send":600,"connect":600},"create_time":1667787147,"name":"domain.com","nodes":{"domain:80":1},"id":"00000000000000001868","update_time":1667787147,"scheme":"http","pass_host":"pass"}

[daoleno]

I encountered the same issue, @wangfeng22 have you found a solution?

• Occasionally encountering 404 route not found error
• Occasionally being routed to a different upstream.