Vhost: Cluster redirect to different vhost or stream

[vladimir131313]

Description'

Please ensure that the markdown structure is maintained.
I am trying to use srs cluster mode to redirect srs-edge to different srs-origins with different vhosts.
For example:
Step One:
client play rtmp -> srs-edge.test.vhost (remote mode) -> srs-balancer.test.vhost (local mode) -> local coworker (this is my application that responds to an API cluster request)
Step Two:
local coworker (my application) Send api answer with Origin: {IP,PORT, vhost=pub1-vhost}-> srs-balancer Send RTMP Redirect with RTMP://IP:PORT/stream?vhost=srs-edge.test.vhost (it ignores vhost=pub1-vhost) -> srs-edge.test.vhost (remote mode) go to rtmp redirect

Please describe the issue you encountered here.
'
Make sure to maintain the markdown structure.

Unfortunately, srs only uses the Host and Port from the api cluster response, but Vhost is still required for my cluster scheme.

1. SRS version: 3.0.140
2. The log of SRS is as follows:
   srs-edge.test.vhost Log

[2020-03-30 19:26:34.459][Trace][61390][354] connect app, tcUrl=rtmp://srs-edge.test.vhost:20135/test, pageUrl=, swfUrl=, schema=rtmp, vhost=srs-edge.test.vhost, port=20135, app=test, args=null
[2020-03-30 19:26:34.459][Trace][61390][354] protocol in.buffer=0, in.ack=0, out.ack=0, in.chunk=128, out.chunk=128
[2020-03-30 19:26:34.561][Trace][61390][354] ignore AMF0/AMF3 command message.
[2020-03-30 19:26:34.642][Trace][61390][354] ignore AMF0/AMF3 command message.
[2020-03-30 19:26:34.642][Trace][61390][354] client identified, type=Play, vhost=srs-edge.test.vhost, app=test, stream=test, param=, duration=-1ms
[2020-03-30 19:26:34.642][Trace][61390][354] connected stream, tcUrl=rtmp://srs-edge.test.vhost:20135/test, pageUrl=, swfUrl=, schema=rtmp, vhost=srs-edge.test.vhost, port=20135, app=test, stream=test, param=, args=null
[2020-03-30 19:26:34.642][Trace][61390][354] source url=srs-edge.test.vhost/test/test, ip=46.242.119.122, cache=1, is_edge=1, source_id=-1[-1]
[2020-03-30 19:26:34.642][Trace][61390][354] create consumer, active=0, queue_size=0.00, jitter=30000000
[2020-03-30 19:26:34.642][Trace][61390][354] ignore disabled exec for vhost=srs-edge.test.vhost
[2020-03-30 19:26:34.642][Trace][61390][354] set fd=11, SO_SNDBUF=87380=>175000, buffer=350ms
[2020-03-30 19:26:34.642][Trace][61390][354] start play smi=0ms, mw_sleep=350, mw_enabled=1, realtime=0, tcp_nodelay=0
[2020-03-30 19:26:34.643][Trace][61390][354] update source_id=355[355]
[2020-03-30 19:26:34.651][Trace][61390][355] complex handshake success.
[2020-03-30 19:26:34.651][Trace][61390][355] protocol in.buffer=0, in.ack=0, out.ack=0, in.chunk=128, out.chunk=128
[2020-03-30 19:26:34.731][Trace][61390][355] connected, version=3.0.140.0, ip=127.0.0.1, pid=55417, id=689, dsu=1
[2020-03-30 19:26:34.731][Trace][61390][355] edge change from 100 to state 101 (pull).
[2020-03-30 19:26:34.771][Warn][61390][355][11] RTMP redirect srs-edge.test.vhost/test/test from 127.0.0.1:30135 to rtmp://10.0.0.2:20135/test/test?vhost=srs-edge.test.vhost
[2020-03-30 19:26:34.801][Warn][61390][355][111] EdgeIngester: Ignore error, code=1018 : connect upstream : edge pull rtmp://10.0.0.2:20135/test/test?vhost=srs-edge.test.vhost failed, cto=5000ms, sto=500ms. : connect : tcp: connect 10.0.0.2:20135 to=5000ms : connect to 10.0.0.2:20135

srs-balancer Log:

[2020-03-30 19:26:34.691][Trace][55417][689] connect app, tcUrl=rtmp://srs-edge.test.vhost:30135/test, pageUrl=, swfUrl=, schema=rtmp, vhost=srs-edge.test.vhost, port=30135, app=test, args=(obj)
[2020-03-30 19:26:34.691][Trace][55417][689] edge-srs ip=151.236.121.5, version=3.0.140, pid=61390, id=355
[2020-03-30 19:26:34.691][Trace][55417][689] protocol in.buffer=0, in.ack=0, out.ack=0, in.chunk=128, out.chunk=128
[2020-03-30 19:26:34.731][Trace][55417][689] client identified, type=Play, vhost=srs-edge.test.vhost, app=test, stream=test, param=?vhost=srs-edge.test.vhost, duration=-1ms
[2020-03-30 19:26:34.731][Trace][55417][689] connected stream, tcUrl=rtmp://srs-edge.test.vhost:30135/test, pageUrl=, swfUrl=, schema=rtmp, vhost=__defaultVhost__, port=30135, app=test, stream=test, param=?vhost=srs-edge.test.vhost, args=(obj)
[2020-03-30 19:26:34.731][Trace][55417][689] source url=/test/test, ip=127.0.0.1, cache=1, is_edge=0, source_id=-1[-1]
[2020-03-30 19:26:34.732][Warn][55417][689][11] use public address as ip: 151.236.121.5
[2020-03-30 19:26:34.732][Trace][55417][689] http: cluster redirect 10.0.0.2:20135 ok, url=http://127.0.0.1:30155/api/v1/clusters?vhost=__defaultVhost__&ip=srs-edge.test.vhost&app=test&stream=test&coworker=127.0.0.1:30155, response={"code":0,"data":{"query":{"ip":"srs-edge.test.vhost","vhost":"srs-edge.test.vhost","app":"test","stream":"test"},"origin":{"ip":"10.0.0.2","port":20135,"vhost":"srs-pub1.test.vhost","api":"srs-pub1.test.vhost:20185","routers":["srs-pub1.test.vhost:20185"]}}}

[2020-03-30 19:26:34.732][Trace][55417][689] rtmp: redirect in cluster, from=srs-edge.test.vhost:30135, target=10.0.0.2:20135, url=http://127.0.0.1:30155/api/v1/clusters?vhost=__defaultVhost__&ip=srs-edge.test.vhost&app=test&stream=test&coworker=127.0.0.1:30155, rurl=rtmp://10.0.0.2:20135/test/test?vhost=srs-edge.test.vhost
[2020-03-30 19:26:34.771][Trace][55417][689] client finished.

My app send:
response={"code":0,"data":{"query":{"ip":"srs-edge.test.vhost","vhost":"srs-edge.test.vhost","app":"test","stream":"test"},"origin":{"ip":"10.0.0.2","port":20135,"vhost":"srs-pub1.test.vhost","api":"srs-pub1.test.vhost:20185","routers":["srs-pub1.test.vhost:20185"]}}}

But RTMP Redirect:
rurl=rtmp://10.0.0.2:20135/test/test?vhost=srs-edge.test.vhost

1. The configuration of SRS is as follows:
   srs-edge.test.vhost

listen              20135;
max_connections     10000;
srs_log_tank        file;
srs_log_file        /var/log/srs/srs.log;
http_api {
    enabled         on;
    listen          20185;
}
http_server {
    enabled         on;
    listen          20180;
}

vhost srs-edge.test.vhost {
    cluster {
        mode remote;
        origin 127.0.0.1:30135;
        debug_srs_upnode    on;
    }
}

srs-balancer.

listen              30135;
pid                 objs/srs-balancer.pid;
max_connections     10000;
srs_log_tank        file;
srs_log_file        /var/log/srs/srs-balancer.log;
http_api {
    enabled         on;
    listen          30185;
}

vhost __defaultVhost__ {
    cluster {
        mode            local;
        origin_cluster  on;
        coworkers       127.0.0.1:30155;
    }
}

Replay

How to replay bug?

Steps to reproduce the bug

1. ffplay rtmp://srs-edge.test.vhost:20135/test/test
2. tail -f /var/log/srs/srs.log
3. tail -f /var/log/srs/srs-balancer.log

Expect

Expected behavior

Please describe your expectation

Describe what you expect to happen

Is it possible to teach SRS to use Vhost Param from API clusters answer to redirect RTMP?
Thanks.

TRANS_BY_GPT3

[winlinvip]

Vhost Alias

This is the Alias feature of vhost, which means alias. When playing different vhosts, they actually point to the same vhost. This can achieve separate scheduling of upstream and downstream vhosts, as well as risk isolation.

Stream Alias

Another type is stream alias, which is an alias for the stream and can be used for hot backup. For example, there are two streams:

• rtmp://xxx/app/streamA
• rtmp://xxx/app/streamB

If the stream is an alias of streamA, stream alias streamA, and if the user plays:

• rtmp://xxx/app/stream

The actual stream being played is streamA. You can use reload to implement stream alias streamB, thus achieving hot backup.

There is a workaround, using FFmpeg to pull streamA out and push it as stream, which is stream alias streamA:

ffmpeg -f flv rtmp://xxx/app/streamA -c copy -f flv rtmp://xxx/app/stream

Please read more about the detail from this post