messaging publisher stuck

[suzhengwei]

I was using multi messaging publishers to send notifications to external collectors. Once connection to one collector refused, it can't send notifications to other collectors. The publisher seems stuck but without useful log.

[Yadnesh Kulkarni (ykulkarn)]

To reproduce this behaviour I added two dummy notifiers in the sink, and a third one(tcp) which actually works, like this
~~~
      publishers:
          - notifier://172.17.1.121:5666/?driver=amqp&topic=cloud1-metering
          - notifier://172.17.1.122:5667/?driver=amqp&topic=cloud1-metering
          - tcp://127.0.0.1:3060
~~~

Even though the first 2 weren't reachable, metrics were being sent to the tcp publisher. Does such a behaviour blocks messaging publisher only ? It would be helpful if you could provide some more details to reproduce it.

Could you also try out using "max_retry" parameter like
~~~
- notifier://[notifier_ip]:[notifier_port]?topic=[topic]&driver=driver&max_retry=100
~~~

[suzhengwei (suzhengwei)]

I didn't use 'max_retry'.

Please try two messaging publishers, as following：
"""
      publishers:
          - notifier://$user:$password@$rabbit_server1_ip:5672?driver=rabbit&topic=cloud1-metering
          - notifier://$user:$password@$rabbit_server2_ip:5672?driver=rabbit&topic=cloud1-metering
"""

If rabbit_server1 not works('ping' with ip is ok, but 'telnet' with port failed), the messages can't publish to rabbit_server2.
#ping $rabbit_server1_ip
PING $rabbit_server1_ip 56(84) bytes of data.
64 bytes from $rabbit_server1_ip: icmp_seq=1 ttl=61 time=0.392 ms
...
#telnet $rabbit_server1_ip 5672
Trying $rabbit_server1_ip. . .
telnet: connect to address $rabbit_server1_ip: Connection refused