Hi all,
We're at the stage with the illumos/SmartOS KPTI (Meltdown mitigation)
work now that we would like to solicit help with testing from the
community. So for those keen to help out with some testing, you can find
links to a platform image with KPTI (and full DEBUG assertions) below.
This is
Jussi Sallinen wrote:
> Yeah, SDC/Triton is opinionated and for example it doesn't involve shared
> storage (eg. vSAN) as primary storage:
> https://github.com/joyent/triton#design-principles
> "A VM's primary storage should be local disk, not over the network -- this
> avoids
On 30/01/2018 2.35, Jeremy wrote:
Does sdc have any HA? For example, if you make a new usb with current
updates, do you need to bring the whole node down with all VMs just to
get the updates, or is there a better approach.
a good way of handling HA on Triton is to use replication
Yeah, SDC/Triton is opinionated and for example it doesn't involve
shared storage (eg. vSAN) as primary storage:
https://github.com/joyent/triton#design-principles
"A VM's primary storage should be local disk, not over the network --
this avoids difficult to debug performance pathologies."
I wouldn't call your thinking "incorrect", but it is a different paradigm
from VMWare/Hyper-V and similar. SDC/Triton is "opinionated" that
HA/Redundancy/etc should be at the application layer, which aligns well
with the increasingly prevailing "modern" design pattern of so-called
Hi Jason,
Thanks for the tip about libdlpi… the existing approach seems a little brittle
so I’ll have a look at this.
As for the vnic protection flags, I’ve definitely been able to save the
“allow_unflitered_promisc” option using vmadm, but it doesn’t seem to take
effect, despite being
For the SoftEther code changes, I’d suggest looking into libdlpi(3DLPI).
It’ll take care of most of the DLPI details for you (including handling
both style 1 and style 2 devices — it sounds like it can’t handle style 1
devices — e.g. /dev/net0), it’s also something that’s been around since
Solaris
I have been looking in the docs for HA config pertaining to compute nodes.
Perhaps it is my thinking which is incorrect, as I am coming from a vmware
way of thinking.
Does sdc have any HA? For example, if you make a new usb with current
updates, do you need to bring the whole node down with all
Good afternoon all,
I've come to a temporary solution to the problem I've been having with
softether:
First, softether uses DLPI to access the network, and there are 2 ways to do
attach to an interface. One involves opening the root node of a network device
(like /dev/bnx) and then attaching