vetal this sounds like the value names are not getting translated...

tsoome: Yes, I think so. Do you think that reason for a ticket ? Also is there any docs how to use it?

ticket is always good

man eeprom is documenting acpi-user-options

I think this whole mechanism needs review, there is too much talk about solaris 10;)

is it possible to integrate inter-zone dns via the zone names? comparable to docker-compose service names? e.g. in docker-compose i can give a container a name (e.g. postgres-service), and via its name it is accessible for other services. e.g. i can use URIs like postgres://postgres-service from another service

id like to cut out handling IP addresses

properguestt: not that I'm aware of

it seems like it could be a neat idea.. though at the same time, being limited to a single physical system seems like it'd limit its usefulness

also, you can give a zone multiple ip addresses... how would that be handled?

in docker-compose you make sure that the containers share a common virtual network

that or you give it a vnic (and often limit the vnic to only allow a specific ip address, which is effectively the same thing)

the host OS could handle dhcp and dns for each virtual network

and for the containers

danmcd / andyf saw the mail about new bhyve sync and uart, would the normal vm’s serial console also use that code? If so, been accessing my vms over serial for a while now without issues

It is the serial console, but... it's used by mdata-client too, and it failed on {Free,Net}BSD guests until jperkin fixed it.

→ OS-8556: bhyve uart is too fast (Resolved)

His tracking down of this is all there on the bug report; pretty instructive.

Interesting, I have bot hit that at all with just using the serial console

But I doubt I hit those rates

jbk: multiple addresses wouldn't be a problem as you can assign multiple A records to a DNS hostname.

jbk: multi-host configs could be dealt with via dns dynamic update

yeah, but most things tend to take the first response (which even if the DNS server is rotating or randomizing the order), tends to cause more problems than it solves

unless it's being used by a client that really is explicitly expecting it

incus (fork of lxd; container daemon on linux) manages dhcp and dns for inter-container communication using virtual networks without the user manually handling any IPs

properguestt: If it's all on one system, and you were itching to get going, I would probably just write out new /etc/hosts files in all the zones haha

very pragmatic solution

is it actually possible to share unix sockets between containers? maybe using bindmount/nullfs?

*zones

The global zone can see the non-global zone file systems.

I guess if you set up some shared loopback mount, then I guess you probably could.