Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

main-noip apps never terminate #19

Open
mor1 opened this issue Aug 11, 2013 · 4 comments
Open

main-noip apps never terminate #19

mor1 opened this issue Aug 11, 2013 · 4 comments
Labels

Comments

@mor1
Copy link

mor1 commented Aug 11, 2013

even if main-noip: ... is specified, mirari generates the backend.ml to create a thread that accepts on a socket, and adds the Backend.run() thread to the join list in main.ml.

this means a unikernel that never invokes networking will never terminate -- the Backend.run() thread remains live in the accept loop forever.

@vbmithr
Copy link

vbmithr commented Aug 13, 2013

As mirage unikernels were supposed to support interface hotplug (eventually), I was fine having an unikernel never terminates. How do you think that should be handled ?

@samoht
Copy link

samoht commented Aug 13, 2013

Maybe the hotplug ability can also be enabled with a flag ?

@avsm
Copy link
Member

avsm commented Aug 13, 2013

I think unikernels should be explicitly modelled as something that is started and stopped. The UNIX version could/should have a CLI to "list running unikernels with IDs" just like Xen kernels do.

This makes management of them much much easier, since I can just start 8 UNIX unikernels on a Linux box and have them use 8 cores in parallel, just as we would under Xen.

I.e. why not go for an explicit hotplug + start/stop model right now, since it's the only eventual model that makes sense. Very Erlang-like...

On 13 Aug 2013, at 11:21, Vincent Bernardoff [email protected] wrote:

As mirage unikernels were supposed to support interface hotplug (eventually), I was fine having an unikernel never terminates. How do you think that should be handled ?


Reply to this email directly or view it on GitHub.

@mor1
Copy link
Author

mor1 commented Aug 13, 2013

that makes sense to me -- making mirari a thing that basically abstracts over xl for both Xen and POSIX unikernels seems like a useful thing. if it can also send hotplug events to model environmental changes, so much the better.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

4 participants