Subject Re: [IBO] Crash
Author lester@lsces.globalnet.co.uk
> The problem isn't Citrix getting ghost processes, but I don't think
> you realise how rare DB server crashes really are, some people go
> their entire careers without seeing a production server crash.

With IB5.6 running, connections that are dropped are not closed. After a
time, the excess connections cause an overflow which crashes the server. Thos
only way to clear the problem is to stop and start the server at regular
intervals - related to the error rate.

IB6 eliminates this problem, and we are NOW running 24/7.

> As for ISDN call your ISDN supplier, and ask them to help you resolve
> the connection problems, they can run tests on the lines and their
> equipment to make sure everything is working on their end.

Fat chance - ISDN is inherently 'unstable'. As it is a dialed connection,
rather than point to point copper, packets are lost. Where network repeaters
are build to use ISDN they include an auto-reconect to cover the problem.
IB5.6 can not cope with these dropouts.

It has taken us 18 months to get ISDN and Frame Relay stable. IBO, IB6 and
local caching works fine. IB5.6 can not be made stable.

--
Lester Caine
-----------------------------
L.S.Caine Electronic Services