Dial-up access to Oracle - SQL hangs reposting

Hi!

Daniel (et al),

Thank you for your reply. Our windows95 PC is using PPP to connect over a phone line from a 56K modem to a Shiva 28.8 Modem Box. Our Network/Server/Client are all IntraNetware (4.11). We don't get any error messages - the SQL statement just hangs. I don't believe an ISDN line is involved (I wish!). I will collect as much more detail as possible from our LAN group to provide you tomorrow.

Further investigation reveals that the offending SQL selects 31 columns from a primary table, and several other columns from 3 additional tables - one of which is an outer join. By trial & error, we found we can run the joins if we select only 28 columns, or select all 31 if we just use the one table. Are we reaching the limit of 28.8?

More generally, is it reasonable to expect this application to run @ 28.8? Our group just got a 3COM box running @56K, but our remote users may not all have 56 at their end.

Again, this same SQL was running fine for 5 months until January. We did crash the server then, as well as add some indexes and even primary keys to some tables. Also, since then, the SQL has run intermittently, and it seemed to help to re-create the indexes/keys as well as restart the instance - but not always.

TIA, Stephen Clarke

Please describe the problem in much more detail... Full version of net running at both ends, protocol etc etc... Errors received would be good aswell... (prob 3113, 3114) My gut feeling is that you may be using an isdn line that is shutting down
while the query is running at the server ...

cheers,
--
Daniel
Oracle UK

Stephen Clarke wrote in message <36E054A7.798BD09D_at_faa.dot.gov>...
>Hi!
>
>Our Oracle 7.3.2.3 Server for Netware installation is being accessed
>remotely over the phone through a Shiva box. Win95 users are hanging
up
>repeatedly but intermittently during a relatively simple select
>statement, whereas WinNT users work fine. One difference I notice is
>that Win95 has to "disable current network connections", whereas NT
does
>not.
>
>The problem seems to have arisen after we added a bunch of indexes to
>several of our tables to decrease other retrieval times.
>Dropping/recreating the particular indexes involved in this select
>sometimes helps, as does shutting down the instance.
>
>Does anyone have experience with this kind of problem?
>
>TIA
>
>Stephen Clarke
>
Received on Thu Mar 11 1999 - 16:22:47 CST