Updating URI of an existing route causes 404 error

[bisakhmondal]

Issue description

If I update the uris of an already created route, necessary effects are getting taken for the old uris name, but for the new ones, it is causing a 404 nginx/openresty error.

Expected behavior

The update should either take place or not, if the manager-api is able to update it should display the upstream content, else the apisix should send an error. But in this case the error is shown from nginx.

How to Reproduce

1. Running all the containers. ./setup.sh up
2. make a put request to managerapi for URI hello & route id r1
   [PUT] http://localhost:9000/apisix/admin/routes/r1

{
 "name": "route1",
 "uri": "/hello",
 "methods": ["GET"],
  "upstream": {
	"type": "roundrobin",
	 "nodes": [{
		"host": "172.16.238.20",
		 "port": 1980,
		"weight": 1
		 }]
	 }
 }

3. Now making an update request with uri newhello keeping the same upstream
   [PUT] http://localhost:9000/apisix/admin/routes/r1

{
 "name": "route1",
 "uri": "/newhello",
 "methods": ["GET"],
  "upstream": {
	"type": "roundrobin",
	 "nodes": [{
		"host": "172.16.238.20",
		 "port": 1980,
		"weight": 1
		 }]
	 }
 }

It disables the old route. but for the updated route it is throwing an Nginx error.

Screenshots

after making the first API call

After updating the uri
The old uri

The new URI

Environment

Running the tests using docker-compose services.

./api/test/docker/setup.sh up

• OS (cmd: uname -a): Linux turing 5.8.0-44-generic 50~20.04.1-Ubuntu SMP Wed Feb 10 21:07:30 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
• apisix-dashboard version, if have: 2.4
• Browser version, if have: Google Chrome Version 88.0.4324.182 (Official Build) (64-bit)

Additional context

[starsz]

The URI in route is /newhello , but the URI in the browser /hellonew.

Please check about it.

[bisakhmondal]

The URI in route is /newhello , but the URI in the browser /hellonew.

Please check about it.

Sorry, my bad. I have mistakenly put yesterday's screenshots. It's updated now. Thank you.

[nic-chen]

@bisakhmondal
what is the content in nginx.conf
and is there error log for this？

[bisakhmondal]

Thanks, @nic-chen, @starsz for looking into this. Actually, the error has nothing to do with manager-api. Thanks to @Jaycean, only /hello, /hello_ and /hello1 routes are whitelisted on the upstream docker image that we are currently using for testing [ ref ]. So in the case of newhello it's a 404 from upstream itself.

The issue isn't relevant anymore to keep it open.

Thanks again, everyone. Closing.