I'm throubleshooting a scenario with a node sever where there are too many ESTABLISHED connections, even when there are no active requests. This from a test environment but same behavior is found on PROD.
This is the number of Established connections on each IP of the test machine:
572 127.0.0.1
48 172.18.0.23
48 172.18.0.16
47 172.18.0.18
47 172.18.0.15
46 172.18.0.24
46 172.18.0.20
31 0.0.0.0
26
2 172.31.45.51
On localhost, Established connections look like this:
tcp 0 0 localhost:59596 localhost:8051 ESTABLISHED hyperledger 478386467 3632665/node /opt/h
tcp 0 0 localhost:44164 localhost:2379 ESTABLISHED etcd 459192509 529/etcd
tcp 0 0 localhost:9050 localhost:48376 ESTABLISHED root 478391433 1653221/docker-prox
tcp 0 0 localhost:8050 localhost:55040 ESTABLISHED root 478390603 1652718/docker-prox
tcp 0 0 localhost:39050 localhost:8050 ESTABLISHED hyperledger 478550015 3632665/node /opt/h
tcp 0 0 localhost:8051 localhost:60804 ESTABLISHED root 478579826 1652946/docker-prox
tcp 0 0 localhost:7051 localhost:44068 ESTABLISHED root 478581029 3355251/docker-prox
tcp 0 0 localhost:37922 localhost:9050 ESTABLISHED hyperledger 478382004 3632665/node /opt/h
tcp 0 0 localhost:53382 localhost:8050 ESTABLISHED hyperledger 478537715 3632665/node /opt/h
tcp 0 0 localhost:45914 localhost:7050 ESTABLISHED hyperledger 478384838 3632665/node /opt/h
tcp 0 0 localhost:9051 localhost:36842 ESTABLISHED root 478388152 1653426/docker-prox
tcp 0 0 localhost:56526 localhost:7050 ESTABLISHED hyperledger 478390799 3632665/node /opt/h
tcp 0 0 localhost:53918 localhost:7051 ESTABLISHED hyperledger 478608229 3632665/node /opt/h
tcp 0 0 localhost:8051 localhost:42296 ESTABLISHED root 478537891 1652946/docker-prox
tcp 0 0 localhost:7050 localhost:56334 ESTABLISHED root 478390037 1652249/docker-prox
...
The localhost connections are half initiated by user root, and the other half by hyperledger user, this is an example with 1 set:
tcp 0 0 localhost:7051 localhost:53946 ESTABLISHED root 478608259 3355251/docker-prox
tcp 0 0 localhost:53946 localhost:7051 ESTABLISHED hyperledger 478609232 3632665/node /opt/h
localhost:53946 -> localhost:7051, this is the node server connecting to a fabric node, and localhost:7051 -> localhost:53946, would be the fabric node running on a container connecting back to node.
As I was saying, this happens even when there are no calls even made. Does this mean that the node server isn't closing the connections? I also don't understand why there are so many node connections running on different Ports. Node webserver is set to lisetn on port 5000, but then uses the fabric SDK to connect to a fabric node on docker. These are the connections and looks like are not closing, even though there is a "gateway.disconnect()" call that should close it.
There are only two node processes:
1716966 ? Sl 17:15 node app/main.js name - hyperledger-explorer
3632665 ? Ssl 7:11 node /opt/hyperledger/vm1/fabric/server/server.js
I think I'm missing some core concepts here, I can't figure out what's wrongs. I'm thinking something with the node app, but couldn't find anything so far and I don't quite get the netstat info.
This seems to be a known issue of the now legacy node SDK fabric-network. For fabric 2.4v onwards, the new fabric-gateway sdk is recommended for improved management of resources.