GitLab

otherwise the file is created as that user, and is then difficult to
remove in floormap_aux.php3 after it is dished out. There is no reason
to run the external script as the user; all it does is build an image.

cached version.

passing this to the newnode client, web page, and viewing and
editing it.

we can use it for a partition number

of my really icky code to start different kinds of servers; basically
you can set the number of servers of each kind, and if a server of
that type exits, another one of the same type is started. The code is
a little easier to follow then it was before, but still quite yucky.

So whats the "right" way? The problem is that if multiple children are
selecting on a set of fd's and an fd comes ready, all of the children
are woken up. This is called a "select storm" since they all wake up,
try to read from the same fd, and all but one go right back to sleep.
Lots of overhead, and programs like apache go to great lengths to
avoid it. Takes a lot code, lots of bookkeeping, lots of goo that is
hard to understand 6 months later. Not worth it for tmcd, at least not
yet. If we start creating many 1000s node experiments, that could
change in a hurry.

format instead of plain old well formed XML. Note this change does not
affect the parser in any way; the parser continues to spit out its own
brand of XML, and xmlconvert treats that special when it reads it.

The XML I spit out now is better formed and makes more sense when you
look at it. It also helps that its much faster to generate, and the
resulting output is much smaller. For example:

	time /usr/testbed/libexec/xmlconvert testbed jail-2440
	real    2m54.229s
	user    1m7.492s
	sys     1m15.927s

	time /usr/testbed/devel/stoller/libexec/xmlconvert testbed jail-2440
	real    0m14.738s
	user    0m5.755s
	sys     0m5.156s

	/tmp/bar*.xml
	-rw-rw-r--  1 stoller  wheel  15595161 Jun  4 11:16 /tmp/rpc.xml
	-rw-rw-r--  1 stoller  wheel   5285690 Jun  4 11:17 /tmp/plain.xml

Thats for a big experiment. For a small experiment:

	time /usr/testbed/libexec/xmlconvert testbed jail-416
	real    0m9.346s
	user    0m5.083s
	sys     0m2.675s

	time /usr/testbed/devel/stoller/libexec/xmlconvert testbed jail-416
	real    0m2.751s
	user    0m1.377s
	sys     0m1.183s

	ll /tmp/bar*.xml
	-rw-rw-r--  1 stoller  wheel  3135518 Jun  4 11:19 /tmp/rpc.xml
	-rw-rw-r--  1 stoller  wheel  1015745 Jun  4 11:19 /tmp/plain.xml

As you can see, the RPC code is rather non-linear in its performance!

Since this affects Tim and netlab-client, I have added an alternate
XMLRPC routine called virtual_topology_xml() that returns the new format,
since Tim will need to transition to this new format (parsing it into a
Java data structure). The old format is left in the RPC server until that
is done.

wireless.

Parse the ptop file to get information like colocation factors, CPU
speeds, trivial bandwidth, and the like, instead of hard coding them.

Keep features and trivial bandwidth by type (ie. pcvm, pcvm850),
instead of making them the same for all types. This should make
much better conglomeration choices.

 allow boot from partitions other than 1
 allow "-new" versions of freebsd/frisbee MFS

line so that we can see the timestamp in it.

graphviz support seems to be broken in at least some versions of
boost with some versions of gcc .

This mostly required messing with the STL #includes.

Still builds under gcc 2.95, and won't be built with 3.3 by default
until I've spent more time testing it.

One reason for doing this is that gcc 3.3 seems to generate faster
code from templated functions. Tests so far show that the
gcc3-compiled binary shaves 15-30% off of assign's runtime.

The other reason for doing this is forward-looking. When we end up
getting boss running on FreeBSD 5 or a recent Linux distro, the
compiler is likely to be from the gcc 3 branch.

/proj right away so that booting nodes see the current copy, not a
stale one (or none at all).

under-counted.

time it successfully mounts a filesystem.  Since we mount about 80 gazillion
filesystems in jails and these mounts are overlapped with other parts of the
setup process, it is quite likely that some jail's innocent use of
/usr/bin/foo is going to be blown out of the water by a random senseless HUP.

Grisly hack: when setting up jails, rename mountd.pid so mount can't find it.
Hah! hup THIS mount!

work on regular, (as opposed to trivial) links too.

that two identical topologies always produce identical topofiles.

and in jails especially.

These get redirected to a file nowadays, so they shouldn't cause problems.

on and off the assign prepass on a per-experiment basis.

No parser interface yet - for use only with manual tweaking for now.

plab nodes.

but now it is based on the topomap file (see tbsetup/gentopofile).
When the router type is static-ddijk, and there is a topomap file
present (gettopmap() returns valid) turn the topomap into input to
feed into the ddijkstra program. Take the output and feed that into
the parsing code.

NB: If we change the output of tmcd routing call, we have to change
the output of this program too. This will likely become a problem at
some point.

present (gettopmap() returns valid) turn the topomap into input to feed
into the genhostsfile program (see initial revision for os/genhostsfile),
Take the output and feed that into the parsing code.

NB: If we change the output of tmcd hostnames call, we have to change the
output of this program too. This will likely become a problem at some
point, although the format of the hostnames command has not changed for
about 2 years now.

The point of doing this is to remove the bottleneck of computing hostsfiles
from tmcd, which centrally computes them for every node, and in a big topo
(100s of nodes) that takes a lot of time and bw.

returns it as a perl data structure to the caller. The caller is
responsible for doing something meaningful with it.

Remove the static-ddijk code I added for the OSDI paper (revision
1.107), since the new topomap stuff supercedes that.