On 22/01/15 16:20, Martin Lucina wrote:
>> Notably, though, the environment will be global for all rump kernel
>> processes running in the same domain (i.e. rump_pub_lwproc_rfork() will
>> not change it).  Now, the issue is: should the environment setting usage
>> have something which will later allow us to fix get/setenv() to be
>> process-wide and specify the env for certain processes.  something like
>> -e 'mysql::env=value' -e 'php::anotherenv=value' (for example).
>
> There's no reason we couldn't do that. However, I still don't fully
> understand the use case for running multiple processes atop a Rump Kernel
> on Xen/KVM. There would be no isolation between them, so either process
> could stomp over all other processes, so your example of mysql and php
> seems particularly dangerous.

Let's not get hung on the php/mysql example.  If putting two processes 
in the same guest is a bad idea per the use case, you of course wouldn't 
do it.  php and mysql just happened to be the first two names that came 
to my mind ;)

> Are you thinking of porting applications which rely on multiple processes
> by design, such as Postfix? I can see that it would make sense there.

I'm thinking of every scenario where it's necessary/useful.  For 
example, if we want to configure a cryptographic disk into a guest, we'd 
need cgdconfig in there, and so forth.

(ok, maybe I should have said that we want RAID, because currently I 
don't know how we'd protect an application vulnerability from leaking 
out the encryption key.  leaking the raid config should be less critical ;)

>> Hmm, was it so that rumprun-posix seeds the guest env with the host env
>> and that there was some reason for it?  Would be nice to have it work
>> the same way as on xen with rumprun (and hopefully on baremetal with
>> rumprun as well).
>
> You mean passing the full host environment as seen by rumprun to the
> xen/KVM guest?
>
> I agree that it would be good to have a common mechanism. However I don't
> know (haven't looked yet though) if eg. KVM has some equivalent of Xenstore
> to communicate between the host and guests. Also, for pure bare
> metal/embedded use cases the interface would likely be the bootloader or
> other ad-hoc mechanism.

No, not mechanism, semantics.

I think it would be cool to be able to sport the same configuration for 
-posix, -xen and -baremetal and have things work in sort of a similar 
way.  Yes, for some embedded IoT-like scenario rumprun would most likely 
produce an image you can flash or whatever, and which has the parameters 
baked into e.g. the bootloader as you suggest.

I still think that we really should shoot for the "develop with 
rumprun-posix, then deploy as you like" model.  Unfortunately, -posix is 
quite complicated and different.

------------------------------------------------------------------------------
New Year. New Location. New Benefits. New Data Center in Ashburn, VA.
GigeNET is offering a free month of service with a new server in Ashburn.
Choose from 2 high performing configs, both with 100TB of bandwidth.
Higher redundancy.Lower latency.Increased capacity.Completely compliant.
http://p.sf.net/sfu/gigenet
_______________________________________________
rumpkernel-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rumpkernel-users

Reply via email to