GitLab

again.

One of the fixes changes the way in which we iterate through pclasses
in find_pnode(). We used to treat the vector like a ring buffer, and
start (randomly) someplace in the middle. This turns out to give some
bad statistical properties when doing dynamic pclasses, since long
chains of disabled pclasses will cause some pclasses to be selected
more often. My old hack of just hopping around randomly in the
disabled-pclass case was bad, because it's hard to tell when you've
actually tried all the pclasses - so, we were getting false negatives
where it was looking like there was no place available where we could
map a vnode, which turned out to have worse effects than I had
thought.

So, now, we make a list of all the indices and randomize the order,
then just iterate through that list.

We also now count the number of pclasses that are enabled at every
temperature step, and adjust the neighborhood size to remove them.
This makes dynamic pclasses quite a bit faster - it cuts the time
by 30% - 50% for my test case.

Cleaned up find_pnode() by removing some #ifdef's that we don't use,
and probably will never want to again - this makes the function almost
readable!

Well, thats the hope. Not sure it will work, but might as well try.

it to the user - it was a source of confusion.

debug problems.

in every routing table except the main before inserting new ones
obtained from tmcc. The problem was that even the interface routes
that were added when the interfaces were brought up are flushed.
Unfortunately these routes weren't getting added again.

The solution is to toggle the interfaces i.e. disabling and re-enabling
them again. It has the desired effect of flushing all the routes as well.
I'm not sure how things worked until now. In all my previous mappings
of the testbed/nse10 straight-line topology, things mysteriously worked!

Fix a bug in enum parsing

Fix typos that were keeping MySQL errors from coming through to the
web interface.

occur if a normal duplex link was created first. The reason is coz the
creation of an rlink would also internally create an IPTap agent and when
this agent was being attached to the node, classifier entries were being
added to the port demultiplexer for all the IP addresses of the simnode.
Unfortunately, if an rlink was created first before anything else, the
IP address would be set only in the next Tcl statement causing nse to
exit prematurely.

not skipped in the case of 10Mb 0ms links.

"upleavel 1 eval $script". The latter is wrong and was detected when a
for loop inside make-simulated did not work.

* use IP addr rather than finickey hostname when communicating with PLC.
* make Node._create() aware of "already assigned" condition.
* Bump vnode_setup timeout back to two minutes (for now).

keys. Previosly, I had based the create definition on the virt_nodes
table. Unfortunately, that turned out to be incorrect.

mechanism in the service sliver).

vnode_setup for the timeout on waiting for child processes.  I've
set it to 10 minutes since all ancillary setup programs have their own
time bounds (I think - the plab ones do anyway).

The function of plabmonitord has changed slightly.  Instead of setting
up and tearing down vnodes, its job is to just setup the emulab management
sliver on plab nodes in hwdown.  Once the vserver comes up and reports isalive,
it moves the node out of hwdown.  Currently, it first tries to tear down the
vserver before reinstantiating it.  In the future, we could get fancier and
try interacting with the service sliver directly before simply tearing it down.

All new plab nodes now start life in hwdown, and must be summoned forth
into production by plabmonitord.

This commit does NOT include support for the node-local httpd.  That will
come soon.

central.  Also, back out Mike's hack, and use the ALLOWED_LIST feature
Austin originally had to limit node scope.

some of the tables such as virt_agents and eventlist are not truly
virtual tables. That is coz they contain the vnode field which is
the same as the vname field in the reserved table. For simulated
nodes, the mapping may change across swapins and the event may have
to be delivered to a different simhost. Of course this problem
isn't there for jail vnodes or pc vnodes.

to utah_elab_svc (slice that Jay created). We determine this by the
rather low tech approach of looking at $USER in the environment!

characters () in them. This is needed in the nse case.

Also adjust timeouts slightly to prevent churning.

reporting.

the log file.

(2^24 max chars) when I made the other changes to libdb.pm and table_regex.

When xmlconvert is called during nseparse, we need to clean up the
physical part so that rows for the new mapping can be added. I have
added this. While I say that there is a physical part, essentially
the vname column has the same value as the vname column in the
reserved table for the PC such as 'simhost-0', 'simhost-1' etc.

from PLC. node_rusage is updated by rusaged script running on the plab
nodes, but that is not installed yet. I'll probably need to tweak
these changes once we start getting some data.

how we know that, so startup might have to be a special case). Runs
every 5 minutes, getting the load average numbers from /proc/loadavg,
and disk usage by running "df" and parsing its output (yuck, I know).
Package that up and send as a UDP tmcc request. Also does the account
update when the reply indicates.

Need to rationalize this with the isalive watchdog at some point.