All,

My old company have a set of databases I set up for them. They now
have a
problem with server periodically locking up and not letting any new
connections.

Info:
1) IB 5.7 (like 5.6 but > 255 concurrent connections).
2) NT4 SP 6, 1 proc, .5 Gig Ram
3) Very simple databases containing images (doc scannings average
35Kb)
4) 17 databases on the server
5) Last one added 21/11/01
6) 10 secondary files per db
7) Average DB = 17 GB
8) Read Only database once file is created (Not in terms of IB 6,
only that
no-one writes to the DB in terms of DML).
9) Can't ascertain how many concurrent users, when the support issue
arises,
IB displayed 30 concurrent users.
10) WinNT states that 6 of the 17 databases were opened at the time.
11) You can connect to the database locally, but remote connections
get
10061.
12) Issue seems to be occuring every week or so now.
13) Nothing in the even logs / server IB logs, server seems fine.
14) Reboot server seems to clear the issue.

Any ideas?

The selects generally are individual selects that include an
individual
select that also gets the document image. My next steps are to:
1) Start logging as many stats as I can for the server and see for any
changes prior to going bang (that may be more difficult than it
sounds).
2) Up the application auditing so that we can see how many selects
occur before the server dies.

In addition to the above, there seem to be some 10055 errors in and
around
the problems. 10055 seems to make reference to full TCP/IP
Buffers......

Finally there are some images that are about 2 - 3 MB in size.

Cheers,

JAC

p.s. I have failed to post via atkins newsgroup mirror - I've had to
come online to do this, am I doing something stupid?