Web server failed: listen tcp 192.168.1.126:59999: bind: cannot assign requested address

[Liguangwei0321]

when run device_simple, get errors:

level=ERROR ts=2023-11-30T03:01:41.415677205Z app=device-simple source=httpserver.go:143 msg="Web server failed: listen tcp 192.168.1.126:59999: bind: cannot assign requested address

[jiekechoo]

You could try run app usingroot .

sudo ./device-simple

Here is a guide you can refer:
https://docs.edgexfoundry.org/3.1/getting-started/Ch-GettingStartedHybrid/

[Liguangwei0321]

i try use "sudo ./device-simple",but it display : level=WARN ts=2023-11-30T07:18:58.141047832Z app=device-simple source=secret.go:111 msg="Retryable failure while creating SecretClient: open /tmp/edgex/secrets/device-simple/secrets-token.json: no such file or directory"

[jiekechoo]

@Liguangwei0321
Run export EDGEX_SECURITY_SECRET_STORE=false in terminal before start the device service. You should run it in insecure mode.
here is a example:

sudo su
export EDGEX_SECURITY_SECRET_STORE=false
./device-simple -cp -d -o

[bnevis-i]

You could try run app using root .

One shouldn't need root to bind any port > 1024. Some paths in /tmp/edgex/secrets/device-simple might need to be re-acled in order to allow the service to read its secret store token outside of docker. Most likely, the bind failure is because there is a process already listening on the port. After this is fixed, I suggest going back and finding out exactly what privilege is neeed. Root processes listening on open ports is hazardous.

[lenny-goodell]

@Liguangwei0321 , this is not an issue with the port, but the IP address. You can not use the host system IP in docker container since that IP doesn't exist in the container. You must use localhost and then expose the port to the host system via ports in the compose file.

[eaton-coreymutter]

If this is Linux and not containerized: you can see if any process is already listening on that port:

sh-5.0# fuser 59880/tcp
516
sh-5.0# ps | grep 516
  516 nobody   1296m S    /usr/bin/core-data

If there isn't anything already listening on that port, I have seen this problem when some other process is using that port as a client. For example, some other service has a connection to core-metadata on port 59881, and got randomly assigned 59999 as the source port, keeping you from using that port as a destination.
To fix that, we do echo 40000-60000 > /proc/sys/net/ipv4/ip_local_reserved_ports as root early in bootup, preventing those ports from getting randomly assigned as source ports.

[lenny-goodell]

Per the error message bind: cannot assign requested address it is an issue with the IP address, not the port.