Hi all,
I just filed (or updated) the bugs found during the BRM install
test, install fest of using AI attended by: Dave (Marker), Evan, Eric,
Jean, Moriah and Sanjay. There were a lot of bugs found:
* Filed bug 6796 - AI Engine is splatting passwords raw into /etc/shadow
* Filed bug 6797 - create-service: Should work okay if DHCP network already
exists
* Updated bug 4599 with comments from Dave Marker
* Filed bug 6798 - Setting up AI server should provide cleaner ways to
configure network
* Updated bug 6252 - /etc/nsswitch.dns issues with AI setup if you install from
livecd
* Filed bug 6799 - Need to document where to find AI image and define what an
AI image is
* Updated bug 6778 as having been hit
* Filed bug 6800 - Need auto-dhcp option for installadm to make an obvious easy
button
* Filed bug 6801 - create-service: Should create-service be called add-image
* Filed bug 6802 - Need a URL for the latest AI image
* Filed bug 6803 - create-service: Doesn't work without a create-client
* Filed bug 6805 - Need to provide check for other DHCP servers on network
* Updated bug 6774 - create-service should take imagepath with the -t option
flag
* Filed bug 6806 - Would like to have distribution of Sys V
SUNWinstalladm-tools, SUNWpython-lxml
* Filed bug 6807 - Generic bug for supporting AI on other OSes
* Filed bug 6808 - create-service: When image dir. already exists dns-sd still
started
* Filed bug 6809 - create-service: Passing non-numeric argument to -i option
errors without grace
* Filed bug 6810 - AI image missing D-Trace toolkit
* Filed bug 6811 - list: should have similar output between list and list -n
* Filed bug 6812 - create-service:Default service names are difficult to
remember and not friendly
* Filed bug 6813 - create-client: Should create-client be called add-client or
manage-client
* Filed bug 6814 - create-client: It is not obvious that a filesystem directory
image is what is desired
Further, there were some questions of architecture raised:
Why use different images or services versus trying to ensure that going
forward we limit backward incompatibilities and the user always runs the
newest image even if installing an older OS version. Then version and
client issues are relegated to manifests which are far more flexible than
client setup which by problem requirements necessitate using MAC
addresses.
To help simplify DHCP setup and client setup, why don't we see if ISC (or
Sun DHCP) can discriminate on Sun owned MAC addresses (IEEE OUI's) to tell
if the machine asking for a DHCP macro is X86 (unknown) or SPARC (Sun
range of OUI's). If X86, return necessary TFTP bits for default service;
if Sun, return necessary WAN-Boot bits. This would be particularly
useful with bug 6800 - Need auto-dhcp option for installadm to make an
obvious easy button.
Thank you,
Clay
