On Sun, Apr 23, 2006 at 10:28:05AM +0200, Guenther Fuchs wrote:
> Hi there,
>
> on Sunday, April 23, 2006 at 9:57:06 AM there was posted:
>
> e> - ability to hot-add/hot-remove IPs from running guest
> This is lost? AFAIK it is still possible, using the proper context.
this is a feature which was _added_ with 2.1 and is
_only_ possible with the new tools, i.e. that is
something which was _gained_ in 2.1 with newer tools
> e> - ability to enter running/non-running guest
> Running guest entering is not lost, non-running guests always started
> them, so I see the actual way beeing more clear
the --rescue option is to 'enter' a 'non-running' guest
> e> - ability to hot-mount directories into running guest
> This is lost? AFAIK it is still possible, using the proper context.
vnamespace -e <xid> -- mount ....
or if you prefer you can disable the private namespaces
completely with the _new_ config :)
> e> - when entering guest this info used to be displayed:
> e> ipv4root is now 127.2.0.1 127.3.0.3
> e> New security context is 10005
> e> it's missing now
> OK, that is lost. But it is not really necessary IMHO. There are other
> and also easy ways to achieve that info.
>
> e> - old vserver-build scripts created skeletons ready to be
> e> deployed, new method requires quite some time for vserver-enabling
> e> newly built skeletons... which is time-consuming, requires some
> e> knowledge and is error-prone.
> Cannot follow this problem, as I never worked / do not work with skeleton.
>
> e> - ability to look at single vserver configuration and understand
> e> what's going on ( similiar task requires running through
> e> directories, checking their contents, and extensive knowledge of
> e> CURRENT way the utils works.) I guess "find . -type f -ls -exec
> e> cat "{}" ';'" would be work-around for this miss-feature.
> I had this as a personal "problem" as well when changing to the new
> scheme. As for now, since I'm working a while with the new scheme, it
> is more clear and more understandable for myself then the old one. But
> there's one point, I agree still with you: The examples are lost, the
> relevant documentation is therefore less understandable (and not
> everyone realises how to switch the stylesheet on the "flower page").
> So this really could (should?) be improved to fit the previousely
> reached level, yes.
>
> e> - way of keeping heavily-commented template config for vserver.
> e> There is no easy way to comment current config.
> Agreed.
what about writing option.info files?
> e> - generally new style trades ease of programming in primitive
> e> languages for sysadmin's time. Which is great for people writing
> e> utils, not so great for those forced to use them.
> ???
I wonder _why_ nobody made a tool to map some kind of
property list (or xml document) to the directory based
layout and back ... seems to me that the admins do not
really care, just use it ...
> e> - principle of least surprise was broken, for example in 'dev',
> > 'nodev',
> ???
well, we already clarified that you have some strange
kind of expectations for that config, which suddenly
differs from the old one (where IPROOTDEV meant exactly
what dev means now)
get over it! stop whining!
best,
Herbert
> --
> regards 'n greez,
>
> Guenther Fuchs
> (aka "muh" and "powerfox")
>
> _______________________________________________
> Vserver mailing list
> Vserver@list.linux-vserver.org
> http://list.linux-vserver.org/mailman/listinfo/vserver
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Tue Apr 25 16:19:41 2006