@Balint, Apologies for not responding sooner.
Perf-wise, the delta between with and without worst-case values from your results: (0.959 - 0.624) = .335s is a non-trivial amount (almost 50% more) overhead for a single connection. Users (or programs) may run concurrent ssh sessions, which I don't see tested here (you said this was out of scope, I disagree). We've also not tested induced failure (you can run an iptable rule to map the IMDS address to be dropped to simulate failure to talk to IMDS) so we don't know what happens in the case that ec2-instance-connect cannot reach the end-point. I think we should understand the behavior in best and worst case scenarios before we enable this. Given that in the images this will be opt-in (disabled by default), I won't block this request any longer. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1835114 Title: [MIR] ec2-instance-connect To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/ec2-instance-connect/+bug/1835114/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs