Artur, Helen, Ann, et al,

I posted a question back in December 01 regarding getting errors 10061 &
10055 after irrelgular periods on a specific server, with very large
databases.

Firstly, thanks for your responses, I certainly have learnt a fair amount
about:
1) MSDN site (i.e. not most user friendly, like an unindexed library, just
keep walking the shelves until you happen to find the book you want)
2) netstat (very useful out of the box utility for windows that shows you
the state of TCP/IP UDP connections)
3) google.com (very powerful search engine)
4) Windows limits (what you don't find written on the box).

I thought that it was only fair that I posted my findings;
1) Windows can only open 180GB of files concurrently. File handling is
dependent on a resource (Paged Kernel Memory), this can not go any higher
than 192000 - On NT you can find this from the "Windows NT Task Manager",
bottom right, see "Paged".

Once you have exhausted this memory, all sorts of scary stuff happens,
random errors, windows not being able to spawn process, windows, launch
EXE's, copy files. This resource depletion was what was causing the 10055 I
think.

So now we have to manage this, I have asked on a forum & specifically
someone from Windows, what the limits on 2K & XP are, but it doesn't seem to
be documented. Anyone know?

I think Peter had a similar problem on this forum testing RC1/RC2 - my
solution is to split the databases in 20GB chunks (which we do anyway), then
build a server side layer that is serves up the document images, it can
cache connections to the database and ensure the 180GB limit is not
exceeded - we need to get this to at least 360GB per server.

2) I have a thought that by combining netstat and client side auditing (all
of our applications audit actions [log-on off] into an audit databases), we
can see exactly who and where people are that have old connections and are
effectively locking records.

Thanks again all.

JAC