fatal multicast disabled error showed up

[zeroecco]

attempting to update a working config in our staging env but started getting a new error from the networking stack I have not seen before (see below):

ops run -n --accel=false auction-gateway
 100% |████████████████████████████████████████|  [0s:0s]
booting /Users/rich/.ops/images/auction-gateway ...
en1: assigned 10.0.2.15
time="2023-03-07T01:03:49Z" level=fatal msg="lo0: multicast disabled"
exit status 3

binary is built on my M1 mac using GOOS and GOARCH set to linux and amd64 respectively:

file auction-gateway
auction-gateway: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), statically linked, Go BuildID=cKSffFqddC7uRfA7fjpc/IByO1FhAaE4qpnovOBW3/GsDzTCYYjrcJywqpclmc/4SaqEFMWwJkKMCBZPSxq, with debug_info, not stripped

Not sure if anything has changed or not at this point to create this issue

[eyberg]

are you actually using multicast? most public clouds won't support multicast out of the box, we could potentially add support if you are deploying this elsewhere (eg: your own machines)

if you aren't trying to use multicast this could be a config setting in whatever you're running that should be disabled or it could be an error but we'd need a snippet of code to look more into it

[zeroecco]

We are prepping to move to a real DC and multicast is part of that work yep! If this is not a new thing in ops/nanos I would like to ask for multicast support for the future. Thank you so much for the fast response!

[eyberg]

ok linking https://github.com/nanovms/ops/issues/1339 - to help prioritize can you expand on your use-case? (is there some underlying library you are using in this auction-gateway we could look at?) also, I noticed this is set for loopback - is this supposed to be talking/sharing with another program locally? if so is it setup to have them be running as multiple unikernels?

[zeroecco]

apologies on this one. I ended up switch jobs and this issue fell off my radar as I no longer have access to this codebase.