Glenn Faden wrote:
> Scott Rotondo wrote:
>>
>> It seems to me that allocate should create the device nodes regardless 
>> of whether a file system was mounted, and therefore a single exit 
>> status for success is sufficient. Even if I mounted a file system at 
>> allocation time, I should be able to unmount it and access the device 
>> node directly without deallocating first. Doesn't that match how it 
>> works in the global zone?

> First of all, the design of device allocation in TX is not the subject 
> of this fast track. TX has always mounted the allocated device without 
> creating a device node in the non-global zone. This case is only about 
> providing better feedback to what is going on so that Sun Ray software 
> can work more reliably.

The design of device allocation isn't part of this case, but it's being 
used to justify the multiple "successful" exit codes, and that is part 
of this case.

Here's a suggestion: Assuming that the device_clean scripts run with 
sufficient privilege to do so, let them create the device nodes. The 
script can always create device nodes, or never, or only when mount 
fails. Or you can implement other unusual policies we haven't thought of 
yet. In any case, you only need one exit status for whatever the script 
thinks is "success" because there is no further action for the device 
allocation code to take.

>>
>> A more basic question: This case mentions the fact that mounting file 
>> systems during allocation is unique to TX, but I believe the 
>> difference is simply due to historical accident. Wouldn't it make 
>> sense to provide this feature to all Solaris users of device 
>> allocation, regardless of whether TX is enabled or not?

> Again, that is not the subject of this case. This is about supporting 
> device allocation in zones by Sun Ray software. If device allocation in 
> standard Solaris is actually important to customers, we could extend the 
> functionality in standard Solaris. However, I think that would done 
> differently; probably based on HAL and the GNOME Removeable Drives and 
> Media application.

As you know, I would very much like to see a grand unified 
implementation for device allocation that works with or without TX and 
plays well with logindevperm, volume management, and Gnome's removable 
media utility. It would require a certain amount of work, but I'm 
convinced it's feasible.

But as you said, that's not this case.

        Scott


Reply via email to