On 12/13/10 12:41, Ginnie Wray wrote:
Hi -
I've added a proposal for converting the ICTs to checkpoints to the
caiman-docs repository.
It can be downloaded from:
http://src.opensolaris.org/source/xref/caiman/caiman-docs/ICT/ict_conversion.pdf
or pulled from the caiman-docs source directly. Instructions are at:
http://repo.opensolaris.org/info/repository.action?projectId=213&repositoryId=128
I would like feedback by Thursday, December 30. If the holidays
interfere with this, just let me know and I'll accomodate you.
Thanks,
ginnie
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
Hi Ginnie,
Here are my comments on the design doc:
- Overall: It would be nice to have a summary of how each of the existing
ICTs are converted in the new CUD world. Perhaps a table or something?
- Section 5, general comment: The __init__() of most checkpoints are
taking some values
as arguments. I do not think this is the right way to do it.
First of all, applications will register checkpoint before they are
executed,
and all the arguments need to be provided at registration time.
Some of the values in the argument list might not be available
at registration time. As checkpoints that will gather these
arguments are executed, they will save the values in the DOC.
So, all the ICT checkpoints should probably look for values in
the DOC instead of requiring them to be passed in at registration time.
- Section 5.1:
* Do we really need this checkpoint? My understanding is that
the sys config project will have a checkpoint that generate many of the
values configured
here as SMF properties which will get consumed at the first boot by
SMF services that will set up the stuff. Among all the things configured
here, I know that all the user related stuff, and the hostname will be
configured
using SMF services. That only left "lang_locale" and "loc_grubmenu".
I think lang_locale will be configured by SMF too, but not sure.
"loc_grubmenu" probably belong in the BootConfiguration checkpoint, since
that's related to booting.
- Section 5.6 and 5.7: Is there a reason to make this 2 separate scripts?
Since they all deal with IPS, does it make sense to combine them?
- Section 5.6, _reset_image_uuid: this one appears to only be used by
installers that does cpio install (LiveCD/text installer). Does it make
sense
to move it to "CleanupLiveCD" checkpoint?
- Section 6, ict_snapshot:
I am not sure whether the snapshots taken by the engine can replace this.
The engine takes snapshots for debugging purposes. If I remember correctly,
if an installation is successful, the installers might call
engine.cleanup_checkpoints(),
and all the snapshots will be removed. Since the ict_snapshot ICT is to
take a snapshot
of the system right after installation time, it serves a different purpose.
Thanks,
--Karen
_______________________________________________
caiman-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
