I asked here before but I guess I can now be a little more precise about
what is happening. In a small LAN with three SBS-Servers on one server
Netware 6.0-SP5, Groupwise 6.0-SP4 (including POA, MTA, GWIA) and Zenworks
3.2-SP3 are running. The server had an uptime of 354 days. A few weeks ago
GWAVA 3.72 was added to the server. After around one week of working fine
on a Monday morning the server was found unable to communicate inside the
lan at all. No ping, nothing. Utilization stood at 99% with GWMTAVS.NLM
being busiest thread and there were of course messages on the console about
"being unable to communicate to server xyz". Besides this and some other
messages about missing communication everything seemed ok. I restarted the
server and everything was fine again. This continued over some weeks with
the server dropping out of the LAN sometime over the weekend.

Today (lucky me) this started during working hours. At this time the
Groupwise-Client (in Online-Mode) is the only real access to the server
(database lies on a different server). In the early morning the GW-Clients
came up slow with building up the client-window-content (list of e-mails
i.e.) even slower. Opening an attachement-file did fail, the GW-Client had
to be stopped via Windows-Taskmanager. It got a little better during
morning and got worse again at noon. The GW-Clients did not start at all.

At this point I had a look on the server via NRM and the server sat there
with utilization around 2% (smiling at me?), all lights green, ping
possible etc.. We had not the time and patience to wait until a total
communication-breakdown might occure, so we restarted the server (and
everything was fine again!). I am very sure, that with some more patience I
would have seen the monday-morning-syndrom.

I had a talk with a GWAVA-guy today and his guess was a problem with the
tcp-ip-stack, but he could not tell where to start digging. Since the
server is able to run for a while it looks, as if something is eating away
something else (like a memory-leak drops cache-buffers).

Does anybody know, where to check and track resources which lack are very
likely the reason for this issue?