GitLab

If the libevent package is installed, there will be /usr/local/include/event.h
which conflicted with our event.h in the clientside/lib/event directory.

things like undefined variables, and print a stack backtrace using
the Carp module.

every now and then. Seems to happen a lot on the racks, and
there are lots of them.

This enhancement uses the new force10_expect module to add/remove
port channels to/from vlans.  Having to do this via expect is
wretched.

going to ignore the listed flag, and show users (guest or logged in) those
profiles they are allowed to instantiate.

is the SA instead of another user.

support the genidesktop.

identity. The credential is just a speaksfor for now, but this might
hold other kinds of credentials in the future.

are different from Emulab.

messes up the stats. Reorg a little to allow for APT and
CloudLab keys.

Not hooked up or tested yet.  Curses be to Force10 Networks that this
module had to be written at all.  There is really no other sane way to
manipulate vlan membership for portchannels (LAGs).  Force10's lame
excuse for a switch-side scripting API just telnets back in via the
loopback and uses Expect itself.  It also lacks functionality that we
require.  Horrific.

'Cos at least one of us is stupid enough to forget this.

From Jeff Bowles <jbowles@newmexicoconsortium.org>.

to handle that since that is one time we do not mind an error.

Fix up LVSIZE calculation to use IMAGESSIZE.

E.g., telling it to save an unused (type 0) partition. Previously, we
would create a single valid range with start at the beginning of disk
(or partition) and a size of zero. However, size zero is taken by the
imagezip compressor to mean "til EOF" so we would wind up creating a
really large image instead of doing nothing.

It is debatable whether this is really an error. Perhaps we should just
create a valid image file with one chunk containing no valid ranges, but
it seems likely that they probably made a mistake and we should call it
out immediately.

A couple of tweaks to debugging output as well.

Probably cause of mfrisbeed assertion failure and crash.

This commit contains enhancements to the snmpit force10 module for
supporting trunk setup.  Brecht Vermeulen from iMinds did the initial
implementation work, and I modified and cleaned it up some.