Intermud-3 router status

[cratylus]

At the risk of jinxing things, I just wanted to do a little
public dance here to celebrate this record.

Total, total jinx!

I spent the last two days carefully testing and re-testing the
new router code on isolated muds, in a "lab environment".

Once I was sure everything was perfect, I slowly rolled it out
into live production as I had practiced, and mirabile dictu,
*yatmim and *i4 survived the upgrade without a hiccup...no
connections lost, and the new functionality in place.

GREAT SUCCESS! Oh with what joy and smugness did I retire to sleep last night.

Then this morning, the building *yatmim is in had a power failure.

Just shows to go ya.

Anyway, it's back up now.

-Crat

[cratylus]

It looks like *yatmim went down this morning. Not sure why.

Lately the building *yatmim is in has been going through
power outages. My wild guess is that with winter approaching,
the campus maintenance folks are testing generators and
whatnot. I say this because these outages were very rare
in the summer and spring.

Anyway, I'm going to jump in my car today and take a ride
down to the *yatmim server room (read: forgotten departmental closet)
and see wtf is going on.

In the meantime, if you can't bear being isolated from your
precious intermud community, you can switch your router connection
to the *i4 router. *i4 is the twin irn node for *yatmim, meaning that
muds connected to one seamlessly communicate with muds connected to
the other. Because *i4 is on a paid hosting network with a hugalargefat
network pipe, it's actually quite a lot faster and more stable
than the *yatmim node. At some point I'll be encouraging folks
to migrate from *yatmim to *i4 anyway, and this might be a good time
for folks to look into that.

The *i4 connection info is:

*i4 204.209.44.3 8080

-Crat

PS I should've kept my fool mouth shut about the uptime. Talk
about tempting the fates.

[cratylus]

Turns out I didn't need to do anything. The *yatmim box and
the i3 server software hadn't gone down at all. Whatever prevented
network access to it wasn't apparent. My guess is some sort
of networking problem on campus that the IT people fixed before
I got to intervene.

Be that as it may, this is a good opportunity to bring to folks'
attention that in general the *i4 node is much more stable
and much faster than *yatmim, and it is a peer to *yatmim, so
you'll be talking with exactly the same muds/channels as before.

Especially if you've run into connection problems with *yatmim
(which is a weak machine on a weak network), you'll find the
robustness of *i4 pleasing.

The connection info is:

*i4 204.209.44.3 8080

-Crat

[cratylus]

I hate to do it but it looks like there are some deficiencies in channel
handling that need to be addressed in the routers. I will be rolling
out changes to the routers late tonight and also tomorrow. You may
be disconnected or experience weirdness. Please forgive any inconvenience.

-Crat

[cratylus]

The *major* overhaul is complete. At first it looked like both nodes were happy, but
I'm sorry to say that *i4 crashed mysteriously about a half hour after its upgrade >:|

Fortunately, *yatmim, its sister node, seems to have survived the upgrade
without crashing, and has thus far maintained its 17 week uptime (woot).

Both nodes are done with the "big stuff" upgrades. There's some tweaking that
will be going on over the next few weeks, but this should not risk disconnections.

Thank you for your kind patience.

-Crat

[cratylus]

Looks like someone decided to see what happens when
you DOS *yatmim, and sure enough, it denies service.

If you're having intermud trouble, you can try switching to *i4 rather
than using *yatmim, which will be down for a little while this
morning while I tidy things up.

*i4 204.209.44.3 8080

-Crat

[cratylus]

*yatmim is back up. I haven't yet had a chance to  thoroughly
investigate what happened, but it appears to involve
someone sending lots of packets doing unusual things.

I'll be poking at the logs and such to figure out what
happened. If I am able to share details, I will do so in
the next few days.

In any case, it was a good excuse to upgrade the
*yatmim computer to a more robust config, so rather
than a wheezing old 300MHz server, it's on an
old-but-still-spry dual 750MHz machine.

I still recommend people get used to the idea of moving to
*i4 permanently. Eventually my old friend on the faculty
will retire, and I'll have no explanation for the university
police as to why I need to be on campus, working on *yatmim.

-Crat

[cratylus]

The server that the *i4 router is on had some trouble this morning,
and rebooted. As far as I can tell, all is ok now, so your intermud
connection should be able to resume.

The neat thing is that the uptime shortly before this happened was:

Router socket daemon uptime: 17w 1d 12h 58m 36s

And this interruption wasn't because of router code, but
rather server stuff beyond my control. Hurray for stable router code!

-Crat

[Tricky]

Stable? Are you willing to bet your reputation on that? 

Tricky

[cratylus]

Stable? Are you willing to bet your reputation on that?

Today a nasty thunderstorm swept through town and *yatmim got
knocked offline for a bit, and in a strange way. It should be up
and ok now.

This is as good a time as any to announce that *yatmim is no longer
the official primary router of the LPMuds.net intermud network. If
your Intermud-3 client is pointing at *yatmim, please take the time
to change the connection to *i4. It is faster, more reliable, and has
more bandwidth.

Please make sure your configuration points to the correct name and
ip and port. The #1 cause of problems when switching routers is
leaving the old name in the name field. This is not good. This is bad.
Change all three:

name: *i4
ip: 204.209.44.3
port: 8080

Let me know if you have problems I can help with.

-Crat

[cratylus]

Hmm...actually it looks like the network *yatmim is on might be a bit screwy.

You may not be able to connect to yatmim for a little while.

All the more reason to switch to *i4



-Crat

[Tricky]

All the more reason for admins to implement alternate routers given in the startup-reply packet.

Tricky

[cratylus]

All the more reason for admins to implement alternate routers given in the startup-reply packet.

This is a good idea and I've added code to the router to
make it so. It is not yet enabled, though.

The reason I hesitate is that Dead Souls is not 100% compliant
with the spec's intent of "disconnect and reconnect to the
preferred router if it differs from the current connection."

http://ebspso.dnsalias.org:6061/intermud3.lpc

I suspect that other libs, written at a time when multiple
routers were more fantasy than reality, are similar.

It's trivial for me to drag the Dead Souls people into
compliance, but I'm wary of dumping a New Way into
the laps of folks running clients from 1996.

-Crat

[cratylus]

So I'm in my car, *fuming*, roaring to campus,
and I'm thinking about all the choice words I'll
have for the slackjawed maintenance guy who
unplugged yatmim without calling me like the
sign says to do, or to catch the idjit student
who thinks he's stealing a computer that will
run his favrit video game, and I notice that
traffic all of a sudden is crazy, and the
traffic lights are out. And the closer I get to
campus, the harsher and more acrid the air
smells, and in the parking lot there are
emergency vehicles and people running around...
so I turned around and drove back, ashamed.

I dunno how long yatmim/dead-souls.net will be down. Let's
just hope everyone's ok.

-Crat

[cratylus]

Apparently the fire wasn't oncampus but elsewhere...the emergency
response was due to folks stuck in elevators...as far as I know
everyone is ok

Also yatmim is back up, as well as dead-souls.net.

Please remember folks that yatmim is no longer the primary
i3 intermud router. Switch to i4 at your earliest convenience.

name: *i4
ip: 204.209.44.3
port: 8080

(make sure you change the name too, not just ip and port. it's important!)