GitLab

code.

Also, as a debugging aid, if you request a ticket for a slice, and
that slice already has a ticket, delete the ticket and generate a new
one.

tcsh as their shell, I've included a bourne shell script called tcsh that
just execs /bin/sh so that these users can log in via ssh or the console.
tcsh isn't necessary and just makes the image larger.

	Usage: cleanupslice [-f] -c | -m | -s <uuid>

where -c is clearinghouse, -m is the component manager, and -s is the
slice authority. Note that -c only makes sense on Utah Emulab since
that is where the clearinghouse is. When using the -s option, the
slice is cleared from the local SA and the clearinghouse by issuing an
RPC call to remove it. For -m, the slice and its slivers are
deallocated.

This script will eventuallt get called from something to handle slice
expiration on the clearinghouse and the slice authority (the current
expire daemon only deals the CM since that is resources get tied up,
so its most important).

We have yet to figure out an approach for expiration at the SA and the
clearinghouse. Never did like that slice renew thing.

(for themself and the slice).  This shouldn't affect anybody else right now,
but is useful for me to test authenticating with delegated credentials.

the proper unique_subject line in it.

sooner, it was a trivial fix.

to numbered update scripts. To install updates:

	boss> cd /your/srcdir/sql/updates
	boss> perl /your/objdir/db/dbupdate tbdb

For developers, please look in the sql/updates/4/1 file for an example
of how to do updates. For each revision in database-create.sql, create
a corresponding script and add it to the repository. The scripts will
be applied in numeric order (see db/dbupdate.in), starting with the
version recorded in the version_info table. This table is updated by
dbupdate as it runs each script.

updates are done.

Quick fix in interfaceDecl/InterfaceSpec to clear up ambiguity and to make InterfaceDecl a mapping of virtual->component rather than just a name.

allowed.

ports into the right vlan.

experiment with just a firewall and no nodes. That was failing
cause the switchs does not create any vlans, and so there is nothing
to trunk with.

a good thing in production.

Add the version_info table for use with future dbupdate.

deletenode.

rc.linux isn't fedora-specific in any way, so I moved it to the linux/ directory instead.

any switches. Happens.

from the vlans table from tbswap. Also put back the test in nfree
that tests for vlan membership, now that snmpit is fixed.

overridden in the .protogeni-config.py file.

doswapin(modify) so that new test in nfree for vlan membership does
not trigger.