Hi Sarah -
These were your outstanding questions. I thought this might be a bit
easier to read.
!. If wanboot.conf is going to be used, and network booting isn?t
required, then we need to specify how other clients utilize wanboot
logging. I think for consumers of logging this information is important.
The approach an application takes to access the web server is
application dependent. The service for remote logging that we're
implementing is web server. For applications that boot over the net,
such as AI, the network connectivity has already been established. So,
we're leveraging that by also using the web server on the install
server. For something like Text Installer, it would establish its
network connectivity and then provide the means for remote logging via
add_handler. This would require that it request an http handler and also
provide a URL for the logger to use to establish remote logging. The
text installer will determine how to
get the url from the user.
The underlying assumption from the logging framework's point of view is
that the network connectivity has been established by the application.
The logging service is providing the framework. The application needs to
establish the network connectivity.
It is apparent that I need to call all of this out clearly in the
document for different consumers and to clarify that wanboot is specific
for AI. It's quite ambiguous on a few of these points.
2. So, maybe I am not understanding something.. and if so, feel free to
tell me :-). The logging service that is going to be setup with
installadm, does that restrict the remote logging location to be the
install server? Can you describe how the client would set the location
of the remote log?
No. The wanboot.conf model would allow you to define a designated
logging server, if you so choose. The only stipulation is that it has to
also be a web server, since that is the model we've chosen for
remote logging. The consumers can pass a url to the http handler and
the logging service will be set up according to the url that is received.
3. I would like us to be sure that we understand there are no security
issues with automatically setting up this logging service. If you could
add this data to the doc when you have it that would be great.
I'll be sure to follow up on that.
4. So, I am not sure you answer answers my question :-). Are you saying
the application instantiates remote logging via the engine, with
add_handler()? If so, that's fine, just not clear to me.
The application decides whether or not it wants to add a remote logging
handle. So, if it does, it would request a remote logging handler.
Because we are using a web server for remote logging, it
would request an http handle and provide a url to the logger. Then, the
logger would attempt to establish the remote logging connection and
report back to the application if the connection failed.
At that point, the application can decide whether or not it wants to
continue w/o remote logging.
Please let me know if this is still muddy.
thx,
--
Ginnie
