#As The Network Turns
11 messages in this thread
Thanks for trying to help. Several member on the forum gave me
various clues, and I was able to stabilize the system. I think the
problem was too much traffic for Lantastic. The projects were very
large and were causing the main workstation to swap up to 10megs per
field. I moved all maps to all machines and no longer have
problems. Hopefull this is the solution, but we'll see in the next
couple of days.
Thanks,
John Tissavary (La Luna cie)
Glad to hear you've got it working. I remember Gary talking about the
problems some networks have with multiple file handling, I wonder if
that's the trouble here? (I had practically no contact with the
network development in R3). You might ask Gary about it when he
returns tomorrow.
There isn't such a thing as too much traffic trouble. Even if it gets
pretty bad, the worse that could happen is too slow everybody down.
It would be like running the thing on floppy disks. You definitely
have something wrong there.
As far as Smartdrv goes, I use it on all computers without any problem
ever. If memory is being messed with, you will see problems but that
has nothing to do with smartdrv itself.
Hmmmm… I wonder why moving my maps to all machines has changed
things. This is too scary. I'll keep plugging and see what happens.
>>There isn't such a thing as too much traffic
In theory I could agree, in practice I'ld say that isn't the case.
When we ran Nwlite as the main network, called files from working
CPUs, it seemed that when the network got busy with traffic (and
slowed to a crawl), the frequency of files getting corrupted
increased. There could have been other factors involved too but when
we switched to Netware 386, the file corruption went away.
I'm pretty sure there were other factors involved. File corruption is
a very serious problem and if the network drivers allow themselves to
do that, they would be pretty much worthless. When too many people
start sending and receiving blocks of data through the network, many
of those packets are lost. The endless retransmissions are the
delays you see. If a packet arrives corrupted, the network driver
will promptly reject it. If a certain packet gets retransmitted too
many times, you would have a DOS error saying so. But under no
circumstance a corrupted file would be allowed to be written.
I'm making this a big point as I see people blaming the Network and
overseen the real cause of the problem. If you switch the network and
the problem goes away, you probably moved the problem someplace else.
If the cause of the problem was the interaction between the network
drivers and some other faulty driver or TSR, nothing can guarantee
you this same driver or TSR won't disturb somebody else down the
road.
I get corrupted files occasionally written across the network. I have
a Lantastic AI 4.1 3 machine network with a non-dedicated server.
Even when I use dos copy /v (for verify). I haven't pinned it down
yet, but if I do a 300 frame render about 1 .tga file will be
corrupted typically. I'm not blaming anyone or anything yet, I just
don't know what's happening.
Martin I experienced this corrupted file with WFW when copying groups
of files in file manager. After isolating the transfers problem to
that involving just one machine I finally found the culprit. The
Sound blaster! Once removed WFW transfered hundreds of TGA's without
a problem. This is strange because I still have a SB in one other
machine and no problems.
For Gus' info. The corruption looked like this: Tga's would have
solid color bands, mostly green, running horizontally. This would
happen to 4-5 files per hundred.I tried this with Lantastic and got
the same results. Anybody looking for a Sound Blaster? <G> This was
different from the corruption I got with smartdrive on large ram
rendering/ paging. That corruption looked like video dropout on
Hi8. White and black horizontal lines to white and green specs
(Pixels).
my problem sounds the same (horizontal lines of noise in the .tga
file), but the cause seems different. I don't use a Sound Blaster (it
was put in my spare parts box a while ago <G>) and I don't use _any_
caching software. The readme.doc in 3dsr3 mentions this problem and
suggests faster network cards. I already have 16-bit 10mb/s cards, so
I don't know if that could be it. I suppose I could try different net
cards, but money's tight and I'll have to wait until my house gets
sold…or that huge contract we all dream of materializes <g>.
Perhaps the TSR was the Nwlite software? What can I tell you? We
worked a Nwlite system for more than a year and the times we saw .TGA
file corruption was only when we pushed the system to the max,
calling .TGA files for Video Post from other machine that were
rendering, and so on.
As soon as I separated out a server for Projects and maps and one for
Frames, the problem reduced at least by 90% then when I went to
Netware 386, it all together vanished.
Mind you I don't consider myself a network expert, just experienced in
what it's taken to keep the Lab working.
I know that and I don't blame you. If you got it to work, I guess
that's all that's important. Novell 3.11 has a lot more mileage than
any other peer to peer network combined. I strongly agree that that's
a much better route. It's just that it's a bit too messy for someone
with only 2 or 3 machines.