GitLab

* Small fix to DBQueryFatal in libdb.py: () is a valid return value,
  don't fail on it(insert/replace); do fail if DBQuery returns None though.

* Fix up libplab.py to not choke on new plab_slices column.

(which changes how we treat violations) is not working as well as we'd
hope on some topologies.

Add hostkey reporting interval to watchdog.

return something!

* incompatible option handling and use removed from gen purpose libs
* Global PLC mutex implemented, but currently disabled
* plabmonitord parallelization cut in half (for now)

I'm still very frustrated with option handling/passing.  Needs more thought,
but the primary issue is that there really isn't a global variable space in
python (global to file, yes, but not global to interpreter invocation).

I've learned that __builtin__ might work for this, but it seems hacky..

* -a: This already existed; it modifies the path given to the server
  so that people without a paperbag shell can run the rpc server.

* -z: Modify the path above so that it runs the server from the
  current user's devel tree instead of the main tree.

* -x <user>: Instead of the current user's devel tree, use another
  devel tree.

So, to run my own server:

	/usr/testbed/devel/stoller/bin/sshxmlrpc_client.py -a -z

To run Rob's devel server:

	/usr/testbed/devel/stoller/bin/sshxmlrpc_client.py -a -z -x ricci

Sorry for all this sillyness.

the defs file when PLABSUPPORT=1 or an error is raised by configure.  Whenever
the official rootball name/version needs do be bumped, do so in defs-default.
You may track the defs-default rootball name in your defs file, or use your
own divergent name for testing.

track them in this file.

are considered part of the emulab service.  I've set it for 'utah_elab_svc'.

reinstall remote tarballs everytime we change our mind.

Also, make sure rusage function processes updates.

client to here.  Policies are things like only plab service slices do
rusage reports and local nodes don't check for cvsup updates, etc.
Maybe these should all be sitevars, but I didn't really want to have
20 different watchdog-related timeout values.

Note: the detection of what is a plab service slice is a hack right
now (string ends in "-20").  Kirk will be giving me a better way RSN.

what I want them to do and properly mirror the perl impl.

way, we don't have to worry about sshtb getting killed but its child
process hanging around.

the webserver. This is not being used anyplace, except to test that it
does indeed work. At some point we may want to change how we talk to
the backend, going through the xmlrpc server instead of direct to
scripts. The reason for doing this would be to eventually split the
webserver off from the DB node, but that will take a lot more work
(basically have to xmlrpc the entire backend, and at present the web
interface uses a lot of scripts)!

Note that mysql 4.0 has builtin support for ssl based connections. We
could use that for access to the DB if/when we want to split off the
web server.

view.

confusion.

Now ps will show info about the request being processed,
or the total number of requests processed when idle.

1. make sure we send an immediate isalive on startup rather than waiting
   for the first interval to pass
2. don't fall back on a tcp call on plab nodes, this just eventually hangs
   all our tmcds on flaky plab machines