Support more than 1024 pods via multiple Linux bridges

[jokerwyt]

For now, Linux hardcodes a limit (1024) for the maximum ports of an eth bridge. The kernel recompilation is needed to make it larger. Therefore, flannel cannot support pods of more than 1024. I think we can use multiple bridges to overcome this problem. We connect them together and assign pods' veth to one of them. Does that solution make sense?

[rbrtbnfgl]

Considering that by default for IPv4 the masklen is /24 for the node CIDR then the number of pods that will be around 250 more or less. Your case requires a specific configuration I don't know if using multiple bridges could be the best solution.

[jokerwyt]

Considering that by default for IPv4 the masklen is /24 for the node CIDR then the number of pods that will be around 250 more or less. Your case requires a specific configuration I don't know if using multiple bridges could be the best solution.

Yeah but putting more than 256 pods in a node is not a rare need according to my experience

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.