You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Upon running a simple python server container listening on a UDP socket with an attached podman network, UDP traffic that is being sent to the port does not arrive.
Versions 5.2.0-dev-5d10f77da and 4.9.4-rhel both were tried with the same results.
This is a MRE of the issue we are having in production. Docker is fine, podman+cni is fine, podman+netavark exhibits this issue. Note restarting our UDP devices or changing the source port is very cumbersome and we wish to avoid this.
There was no difference from running nping on localhost versus running it on a different machine that can access the podman container
Additional information
Starting the python server first and then starting the UDP sender works as expected but this doesn't help our use case.
Stopping and restarting the UDP sender program while the container is running doesn't help. Only by changing the source port of the UDP sender program does traffic start being received, but we cannot easily change the source port of the UDP traffic.
The text was updated successfully, but these errors were encountered:
This is likely because we do not change any conntack entries in netavark. We must call into the kernel netlink API to drop the stale entries and last I check our netlink did not have any support for conntack types so we would need to implement the types from scratch which is a lot of work.
In any case this is a netavark issue so I move it there.
Note if you are RHEL user it is best to report this through the Red Hat support channels so this can get better prioritized.
Issue Description
Upon running a simple python server container listening on a UDP socket with an attached podman network, UDP traffic that is being sent to the port does not arrive.
Versions 5.2.0-dev-5d10f77da and 4.9.4-rhel both were tried with the same results.
This is a MRE of the issue we are having in production. Docker is fine, podman+cni is fine, podman+netavark exhibits this issue. Note restarting our UDP devices or changing the source port is very cumbersome and we wish to avoid this.
Steps to reproduce the issue
Steps to reproduce the issue
podman build . -t podman_udp_test
podman network create podman_udp
nping -g 17580 -p 17000 -c 1000000 --udp 127.0.0.1
podman run -p 17000:17000/udp --net podman_udp_network podman_udp_testing
Describe the results you received
No output
Describe the results you expected
Output from the server after receiving packets
podman info output
Podman in a container
No
Privileged Or Rootless
Privileged
Upstream Latest Release
Yes
Additional environment details
There was no difference from running nping on localhost versus running it on a different machine that can access the podman container
Additional information
Starting the python server first and then starting the UDP sender works as expected but this doesn't help our use case.
Stopping and restarting the UDP sender program while the container is running doesn't help. Only by changing the source port of the UDP sender program does traffic start being received, but we cannot easily change the source port of the UDP traffic.
The text was updated successfully, but these errors were encountered: