Networking Problem?

  • Thread starter Thread starter D. Eilenberger
  • Start date Start date
D

D. Eilenberger

Sorry if you saw this message on other newsgroups - but
I'm getting desperate and thought it might best be posted
wherever I think the answer might be..

Background -

I moved a consulting client from a NT4 Server (running on
an old 200MHz box) with W98 and XP-Home client machines to
a full W2K-Server environment with all W2K-Pro client
machines.

Since then - I've had nothing but trouble - especially
with an Access application that they rely on to run their
business.

One problem I have - is I am not near the machines when
the multiple users are accessing the files - it's a
consulting job being done out of hours..

The current Access problem seems to be random disconnects
of the Access data files from the client machines.

I spent some time with the systems last night, and was
able to see the disconnects.

The program is an Access one - with one main file using
other files (about 5) with various tables in them. The
main file uses pointers (I won't claim to be at all expert
on Access) to the other files, doing various lookups on
them, and opening and closing them. The pointers are setup
to point to the files using a mapped drive (N:)

The symptoms are simple - about every 15-25 minutes the
user will have the application freeze, followed by the
generic warning message "Disk or Network Error". They
frequently see this message 5 times - which to me
indicates they're getting the warning for all 5 files in
the application.

If they can manage to kill the application (sometimes it's
necessary to use Task Manager) - they can very shortly go
right back into it.

Originally - the main program (file) was located on each
users machine - so they could customize their view of the
program to their needs.

Last night I tried setting everyone up to use a version of
the main program that is located on the server - no joy
here - they called this AM and reported the same problem.

I did not redo the pointers in the main program to use UNC
paths - they remain the mapped drive on the user's machine.

The user this AM who called - after clicking on the
warning message (which appears about 5 times before
letting them out of the program) - then reported the
following message appeared:

"The Microsoft Jet Database Engine cannot find the input
table or query RMA - make sure it exists and the name is
spelled correctly" - I have to assume this is just a
warning from the JetDB engine that it also lost
connectivity to the table "RMA" in one of their data files.

I went through the server last night - and checked some of
the server settings - setting the default inactivity
timeout to infinity (well - 8,171 years according to MS,
close enough) and checked the error log.

There were some NetDDE errors - but these appeared related
to the service for Microsoft Time - and they stopped when
I disabled MS-Time. No errors appear in any of the logs
corresponding to when the Access failures occur.

It doesn't appear immediately to me to be a real network
problem since the users can access other files on the
server without any problem..

The main server is W2K-Server, Patch-3, the JetDatabase
engine was updated with the latest patch from MS, the
server is running dual 1.8GHz processors, with a raid-5
array, and dual (teamed) network cards. It is a Compaq
(forget the exact model - but current one) Proliant (570?)

It is the primary (and only) domain server, running active
directory and DNS. The network used to use DHCP assigned
IP's from a router, I've now set it up to use mapped IP
static numbers on all the machines outside the range the
router DHCP server provided. I made this change recently
due to other problems with trust relationships that were
occuring... and it's possible this might have something to
do with the problem. Before the trust relationship problem
occured - they could use the application. There is a
noticeable speed increase in the application since the
static IP numbers were put in place - but I don't know if
this has anything to do with the problem. All the machines
show up in the DNS widget now - where before they didn't.

I'm giving thought now to resurrecting the NT4 server
until I can solve this problem - and moving the program
back to it.. the user isn't going to tolerate the behavior
they're seeing for very long. It's a bodge to do this, but
it might be necessary.

Any ideas? Should I be taking this to another newsgroup?

Thanks for ANY help! And sorry this is kinda long - but I
thought more details might trigger some thoughts.

Don
 
Got WINS installed on the server? W9x needs it.
Upgrade the XP-home to Pro. The home-version is not able to join a domain.

Let the server do the DHCP and DNS.
How are you connecting to the internet? How many nics in the server and how
are they setup?

Marina
 
Marina - great answer, but not to the original question,
which was mine..

I'll snip out the extranious stuff and leave my question..

Don
 
You can't join a XP-home to a domain. XP-Pro can.
XP-Home can work in a workgroup-environment though.
Make sure all computers are in the same workgroup and all useraccounts exist
on all computers.

Marina
 
-----Original Message-----
Still, your problems could be the result of the issues I
mentioned.

Marina, thanks..

Possibly, but I'll reply to the issues below..

As I noted - we are not using DHCP anymore on this LAN -
it's small enough (15 machines) that it was more trouble
than it was worth... so ALL the IP numbers are now static
assigned ones - with the result - the network access seems
MUCH faster.

ALL the machines are a version of W2K - the server is W2K
Server, and the client machines are W2K-Pro. There are no
XP or W98 machines on the network, so WINS shouldn't be
needed.

The server has dual network cards - setup as a "team"
using Compaq's teaming tool. The server is much more
powerful than the NT4 server the files used to reside on,
it's a current model Proliant with Dual Processors and
heaps of memory (I believe I spec'd 1GB of main memory)
and fast SCSI Raid-5 array for drives.

I can't see server speed being the problem - typical load
is below 1-2%..

So, we're back to the questions below..

=========================================================

Sorry if you saw this message on other newsgroups - but
I'm getting desperate and thought it might best be posted
wherever I think the answer might be..

Background -

I moved a consulting client from a NT4 Server (running on
an old 200MHz box) with W98 and XP-Home client machines to
a full W2K-Server environment with all W2K-Pro client
machines.

Since then - I've had nothing but trouble - especially
with an Access application that they rely on to run their
business.

One problem I have - is I am not near the machines when
the multiple users are accessing the files - it's a
consulting job being done out of hours..

The current Access problem seems to be random
disconnects of the Access data files from the client
machines.

I spent some time with the systems last night, and was
able to see the disconnects.

The program is an Access one - with one main file using
other files (about 5) with various tables in them. The
main file uses pointers (I won't claim to be at all
expert on Access) to the other files, doing various
lookups on them, and opening and closing them. The
pointers are
setup to point to the files using a mapped drive (N:)

The symptoms are simple - about every 15-25 minutes the
user will have the application freeze, followed by the
generic warning message "Disk or Network Error". They
frequently see this message 5 times - which to me
indicates they're getting the warning for all 5 files in
the application.

If they can manage to kill the application (sometimes
it's necessary to use Task Manager) - they can very
shortly go right back into it.

Originally - the main program (file) was located on each
users machine - so they could customize their view of
the program to their needs.

Last night I tried setting everyone up to use a version of
the main program that is located on the server - no joy
here - they called this AM and reported the same problem.

I did not redo the pointers in the main program to use UNC
paths - they remain the mapped drive on the user's machine.

The user this AM who called - after clicking on the
warning message (which appears about 5 times before
letting them out of the program) - then reported the
following message appeared:

"The Microsoft Jet Database Engine cannot find the input
table or query RMA - make sure it exists and the name is
spelled correctly" - I have to assume this is just a
warning from the JetDB engine that it also lost
connectivity to the table "RMA" in one of their data
files.

I went through the server last night - and checked some of
the server settings - setting the default inactivity
timeout to infinity (well - 8,171 years according to
MS,close enough) and checked the error log.

There were some NetDDE errors - but these appeared
related to the service for Microsoft Time - and they
stopped when I disabled MS-Time. No errors appear in any
of the logs corresponding to when the Access failures
occur.

It doesn't appear immediately to me to be a real network
problem since the users can access other files on the
server without any problem..

The main server is W2K-Server, Patch-3, the JetDatabase
engine was updated with the latest patch from MS, the
server is running dual 1.8GHz processors, with a raid-5
array, and dual (teamed) network cards. It is a Compaq
(forget the exact model - but current one) Proliant
(570?)

It is the primary (and only) domain server, running
active directory and DNS. The network used to use DHCP
assigned IP's from a router, I've now set it up to use
mapped IP static numbers on all the machines outside the
range the router DHCP server provided. I made this change
recently due to other problems with trust relationships
that were occuring... and it's possible this might have
something to do with the problem. Before the trust
relationship problem occured - they could use the
application. There is a noticeable speed increase in the
application since the static IP numbers were put in place -
but I don't know if this has anything to do with the
problem. All the
machines show up in the DNS widget now - where before they
didn't.

I'm giving thought now to resurrecting the NT4 server
until I can solve this problem - and moving the program
back to it.. the user isn't going to tolerate the
behavior they're seeing for very long. It's a bodge to do
this, but it might be necessary.

Any ideas? Should I be taking this to another newsgroup?

Thanks for ANY help! And sorry this is kinda long - but
I thought more details might trigger some thoughts.

=========================================================
Don
 
-----Original Message-----
Do you have DNS installed on the server? Does an IPCONFIG /All from the
client give the result that everything is pointing to your server-IP (except
from the static IP you gave the client)?

Marina

Both DNS and ActiveDirectory are on the server.. and the
DNS on the server does see all the clients..

I'll check IPCONFIG /ALL tonight..

I have checked all the clients with NSLOOKUP - and they
can find the server (using the server's DNS) and the other
clients plus outside IP addresses..

What are your thoughts here? (Since I'm heading out there
tonight - and the problem did start to occur when I moved
the client machines to static IP addresses..) [the entire
ActiveDirectory scenerio and DNS are making me start to
think that NT4 was more suited for a small office
enivironment..]

Best,

Don
 
Back
Top