You have this problem because tftp is delivered as a legacy inetd.conf  
entry to preserve compatibility with the setup_install_server script  
supplied for Jumpstart services and thus has no presence in SMF until  
it's added/uncommented to inetd.conf and inetconv is run.  Is there a  
reason you can't just fold its absence into the test as an expected  
case?


Dave

(brevity or typos may be blamed on the iPhone)

On Oct 16, 2009, at 3:23 PM, Andre Molyneux <Andre.Molyneux at Sun.COM>  
wrote:

> I've run into an issue with the installadm test suite when run on a
> newly-installed server that has never had an x86 AI service set up
> on it.  The first test case run for 'installadm create-service' is
> a negative case that provides no arguments.  It expects that the
> command will fail and that mdns, tftp, and dhcp services will
> remain in the disabled state.
>
> The problem I'm running into is with the status of the tftp service.
> Based on the messages from the test suite, what I think is happening
> is that when 'svcs -H svc:/network/tftp/udp6:default' is run the
> first time (before a successful x86 service configuration), instead
> of getting
>
>    disabled <timestamp> svc:/network/tftp/udp6:default
>
> printed to stdout as we expect, nothing gets printed to stdout and:
>
>    svcs: Pattern 'svc:/network/tftp/udp6:default' doesn't match any  
> instances
>
> is getting printed to stderr.
>
> This only happens following initial installation.  Once an x86 service
> has been configured once, all subsequent calls to 'svcs -H' get the
> 'disabled' message, even following reboots.  So this is a one-time
> problem right after initial system installation.  Unfortunately, I
> expect this will bite anyone like ON-PIT who attempts to run the
> suite.
>
> Changing the order of the tests (so that a successful x86 service
> configuration happens before this test gets run) isn't a full
> solution as individual tests can be run standalone and therefore
> still hit the problem.  I can add code that will consider the
> above error output to be equivalent to 'disabled' or 'offline',
> but that could mask problems with the service itself.
>
> Is there any way I could get tftp status to show up in SMF without
> having created an x86 install service on the machine?  If so, I
> could add that to the 'startup' function to work around this issue.
>
> Regards,
>
> Andre
> _______________________________________________
> caiman-discuss mailing list
> caiman-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss

Reply via email to