GitLab

Merge branch 'master' of ccutler@git-public.flux.utah.edu:/flux/git/users/mike/emulab-devel into tpm-tmcd

Cody updated his year old elabinelab and uncovered some problems with the
update process.  Note the updates to the install/updates scripts are just
to add "MASTER_SITE_FREEBSD=1" to the make lines.  This doesn't change
anything functionally, it just speeds up the builds for older FreeBSDs where
some of the MASTER_SITEs no longer exist.

We had a special hack to make sure general/testbed_shutdown sitevar exists,
however we were trying to initialize the ns_include column that doesn't
exist til DB update 191.  Now we just explicitly set known-to-exist columns.

SA of the caller. This means we call out to the Emulab newgroup script
to create a group if it does not exist. We also have to reset the
groups of the GeniUser once we know the SA. 

Other small changes:

* Minor module shuffling.
* Use GetSiteVar() from emutil instead of libdb
* Get rid of obsolete code referring to the old reservation holding
  experiment.
* Cleanup logfiles in the workdir.
* Capture output from the mapper and reformat error logging to make it
  more apparent to the caller what went wrong.
* Change call to GeniSliver::Node->Create(); pass in the node object.

the of the virtual nodes on the vhost are gone. We were hanging onto
vhosts for longer then necessary.

look for an existing credential in the table.

Clean up GeniSliver::Node::Create(); remove duplicate code and unify
the virtnode vs pnode cases. Also change the arguments; instead of a
uuid for the node, just pass in the actual node object from the CM.

for logging errors.

CM to capture the output of varuous commands so that the logs can be
sent to the user, with formatting.

Add subgroup handing in FlipToGeniUser(), and also add ResetGroups()
to change the user's group membership, since the slice experiments are
now created in a subgroup of GeniSlices (named by the user's SA).

toString() that says how to indent the string. Very useful for
printing debugging output. Defaults to 0 (no formating).

an email message. 2) On real errors, stick the libaudit output into
the "output" slot of the GeniResponse structure.

subgroups of GeniSlices, so need to return that extra group
and group membership for users bound to the slice.

want it in the repo.

might not be able to write the output file. Better to us syslog anyway
since its a low usage command.

Conflicts:
	db/libdb.pm.in
	event/stated/stated.in

Also did some cleanup.

Mail follows:

From: Keith Sklower <sklower@vangogh.CS.Berkeley.EDU>
To: ricci@flux.utah.edu
Subject: bug in snmpit_hp.pm
Date: Sat, 6 Nov 2010 13:15:46 -0700 (PDT)

There are two extra characters in snmpit_hp.pm that shouldn't be there.

For some reason, it doesn't seem to matter to older firmware running
on our 5412s that I wanted to do something with a non-existant port,
but a lowly 1U 2810 that we were testing out as a control switch
threw a hissy-fit.

Conflicts:
	db/libdb.pm.in

Conflict resolved by using libdb.pm.in from master, and moving new
constants committed on the tpm-tmcd branch to the new
EmulabConstants.pm.in file.

operational and may never be so. Just want to get it into the
repo.