Re: [Testbed-admins] os_load hangs on new testbed

[Leigh Stoller <stoller@flux.utah.edu>]

> On a newly (semi) deployed testbed we're having the os_load command
> time out.  Machines can have images unzipped to them, and can be
> reached over the console, but just sit waiting when told to os_load
> and eventually the command times out around 12-15 minutes if I
> remember.  The machines do find boss correctly, and can obviously
> boot to the admin MFS without trouble.  But, they just sit
> indefinitely when given a command to os_load.

Hi. Since these machines have consoles, can you please send us the
console output? If that is too difficult, can you tell us if you get
something like this:

> Running /etc/testbed/frisbee -S 155.98.32.70 -M 194   -i
>       155.98.36.38 -m 234.11.138.59 -p 6183 /dev/ad0
> Bound to port 6183
> Using Multicast
> Joined the team after 0 sec. ID is 1779907174. File is 593 chunks
> ....

If you get the first line, but not the "joined the team" line, then
you might be having multicast problems on your control network. Has
frisbee ever worked for you?

Lbs
ps: You are sending email from @gmail.com now, but your email address
on testbed-admins is from isis?