Lantastic 5 Lock-up
12 messages in this thread
After raving about how easy and reliable Lantastic has been for me, I recently
began having some problems:
-> One of my machines locks up giving me that Lantastic "be-deep" sound
indicating it can't read or write. It freezes up, leaving the current frame on
the screen. It does not allow the other system to write information to it
across the net either.
My present rendering patch consists of two 486-66's attached to the end of a 5
system coaxial ethernet. The other 3 systems typically do not get called in to
render as they lack sufficient RAM for most projects: their function is more
for administrative support.
System "A" contains the 3DS network directory, redirected on both systems to be
drive "T:". Both systems dump rendered frames to drive "I:", which is
physically a directory on System "B". The project is configured so that all
maps are read from local drives. The only network activity should be obtaining
the next frame to render from "T:" and dumping the rendered frame to "I:".
Any recommendations or suggestions would be greatly appreciated.
Dave Marks
(203)495-8612 (work)
David you described your system but what have you done different lately that
triggered this lockup? You give the impression that this problem just started
to show up. I get the "bd-deep_ warning everytime I shut down a machine that
has a redirector on the server even if it is not in use.
>> David you described your system but what have you done different lately
>> that triggered this lockup?
I had an important project to render! <g>
Seriously, though, I don't think I ever had the network perfectly bug-free. I
strongly suspect it's related to a higher volume of network traffic. I do tend
to let problems go, and in the past I had occassional lock-ups, but never
consistant enough to warrant the time it would take to run it down. Most of
the time it would run fine, and I assumed the lock-ups were just a freak thing.
It seems to be a little more common lately. My last two attempts to net render
only made it about 100 frames before locking. I've got a project coming due,
so I can't take chances now.
The funny thing is, I have had one machine in slave mode all day with no
problems. It's only after the second machine goes into slave mode that the
problems start. (Although wouldn't you know it… 300 frames ran fine this
afternoon!)
I will be perusing the Lantastic docs, and some notes that I saved from the
forum. In the mean time, I've got a good project to render over the next
several days, and I'll see what happens!
Thanks, Don.
– Dave
David –
We are running 8 486's (plus 3 support computers) under Lantastic and have
had an epidemic of "be-deep's" for the past few months. As best as I can
figure, I can trace the problem to three 486-66 clones we aquired at the
beginning of the year. The problem seemed to happen when the 66's were
accessing their local drives either starting 3DS, quiting 3DS, or saving an
image. If during this "vulnerable" state, someone accessed the drive from
across the network – BE-DEEP!!!
The problem seems to stem from either hardware problems (read cheap
computers) or increased network traffic (more likely). It seems that the three
66's pushed us over the magic 5-6 machine limit. Without moving to a more $$
network, I have made the following changes which seem to work for now. First I
moved the network control computer to a non-rendering machine, then made
everyone strip out all non-local map paths and copied all maps locally.
Secondly I created boot options for server and workstation modes. Thus when net
rendering I make everyone re-boot into workstation mode before putting a
machine into slave. This seems to isolate a machine from all of the network
chatter. This last step has reduced slave-mode blow-ups quite a bit.
One related note. We still had the problem of blow-ups during the day when
everyone was working in server mode (the office's style requires that we use a
peer to peer network). To combat this, I moved all project related files to a
central location. I have had a few grumblings about slowing down file
transfers, but none about blow-ups (my fingers are still crossed).
Hope this helps out.
– Peter Noldt, Archimage
>> epidemic of "be-deep's"
Very excellent suggestions. Thank you very much. I'll implement many of your
recommendations this afternoon. We suffer from both inexpensive hardware and
increased traffic from running 90% peer-to-peer. I'll return as many
workstations as possible to "workstation" mode.
It just occurs to me that our office manager just brought another "server"
on-line! Ah ha!
David –
"Be-Deep" is the most dreaded sound in the office. We have an open plan with
six people. When you hear "Be-Deep…..Be-Deep…." you look around to see
whose face is the redest. The punishment……buy everyone lunch.
Seriously, good luck sorting things out, and please let me know how it turns
out. Our network setup is a work in progress, and by no means the answer. I am
always looking for a "better" solution.
– Peter, Archimage
>> "Be-Deep"
Lantastic has overall been a great convenience. I love the peer-to-peer
benefits. I've had Novell experts in awe with the flexibility. However, the
dreaded "be-deep" can often mean the loss of a night's rendering time.
The built-in flexibility means that things here are constantly changing to meet
our needs. For instance, 2 weeks ago, the administrative assistant (notice the
PC language?!? <G>) got a 500 MB hard drive. Now everyone on the net wants to
grab up that storage space!
After thinking through the common sense advice you gave, I realized another
"little" change that I recently made. At first glance, it almost makes sense,
but on further reflection, I realize the mistake in my logic.
To "share the work", I put the 3ds network files on system "A" and stored the
images on system "B". This means that both systems need to access the net. If
the net files and images are on the SAME system, only ONE needs to access the
net. Whammo! No more collisions! I'll be trying some changes today, and
we'll see how things work out.
Thanks again. (and thanks also to DonL, who got me thinking about recent
changes).
– Dave
Dave –
To describe our system in your terms, if systems "A" and "B" are the rendering
stations, both would be in workstation mode logged only into "C" (System "C" is
a 386 that holds the 3DS network control files). All maps are copied onto "A" &
"B" and all rendered images are saved locally (i.e. "A" renders to "A",
etc…). Thus the only network traffic is A -> C & B -> C. Also A & B are
effectively "locked out" unless rebooted to server mode.
When it comes time to lay to tape, all machines become servers, and thus
accessible to our Targa+/Diaquest setup.
Sorry if this is redundant, but re-reading the thread made me realize how your
system is set-up.
– Peter, Archimage
Thanks for the thanks. <G>
What I do here is to have a subdirectory on each slave: C:\Maps I send all
current maps to each slaves HD before starting to render. It is also listed in
the map path on each slaves 3DS.set. I also list (as a backup) my server's
maps directory on each slave. Of course the IPASes are on each slave too.
I had a problem with lantastic 5 and 3ds the other day. I have a 486 and a P60
and I just started a net render when the 486 failed. I remembered I had
forgottten to copy the maps to the 486 so I shelled out of 3ds and copyed the
maps off the CD to the 486 and Blamo the 486 locked up tight I had to power
down to reboot. Has this ever happened to you ? I havent tryed it again.
>>…and Blamo the 486 locked up tight I had to power down to reboot. Has this
ever happened to you ? I havent tryed it again.
Why not? What are you afraid of?
Yes, once I lost the little key to my 486 keyboard and everything was locked up
tight. I looked all over for it and when I stood up I noticed something very
painful under my foot. The key must have fallen off the table and landed in my
boot. I too had to reboot to get the key out to unlock my 486 but it didn't
require a major shutdown as in your case. Of course I don't wear a size P90
either. 🙂 🙂 🙂
Hey, what kind of response do you expect at 3:30 AM?
Seriously, it sounds like a fluke of unrelated events and … well… welcome
to the world of PCs. Or, as GY would say you were just having a bad day.
That might have been a fluke but I have had a strange case of 3ds.exe getting
wacked out. the problem is only noticed in the Lofter when I zoom extent I get
a glitch across the screen then the pointer arrow goes and the keyboard only
beeps
again I have to power down. This is becoming a problem the other day I lost
about an hours work because of it. I thought it might have something to do with
lantastic and WGFW or Qemm. so I have started to clean out the config and
Aexe.bat . have you ever had or heard of this occurrence?
>Hey, what kind of response do you expect at 3:30 AM?<
are not all good little animators up a 3:30am?