Running a small cluster on EL9, I encounter some kind of bug since 5.14.0-427.31.1 : allocations can’t reach their own exposed port (eg, if a port 22000 is expoed on the host 10.99.7.10, the allocation for which this port is exposed can’t reach 10.99.7.10:22000, there’s just a timeout with no trace of a dropped packet, just no response). Before this version everything was working correctly. But now, the only work arround I have is to disable bridge-nf-call-iptables. Nomad’s documentation states this should be enabled. What do I loose by disabling it ?
For anyone else having this issue : I solved it by enabling bridge_network_hairpin_mode in nomad client’s conf
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.