Revisiting this as we have a number of machines now giving the above error....

Servers = NW65SP8 IP only
Workstations = WinXP SP3 > Client 4.91SP5 + fixes, IP only

Multiple sites. Single Scope defined.
SLPDA on three servers on main sites (20Mb/s links)
Each SLPDA has a R/W replica of the SLP container.
All servers register to this scope and the above 3 DAs
DAs cross register to other 2 DAs - don't register to themselves.
Services appear to register in SLP and hence eDir without issue.

Workstations get SLP settings via DHCP - but we also populated a single DA into the Client (to assist with VPN connections).

Network switches have Spanning Tree off.

Client Setting - SLP Resolve Nearest Server = OFF (fixed the previous issues with same error)

We have a HIGHLY mobile workforce and DYNAMIC infrastructure changes so I do not want to specify a server in the client - I would like to fix our SLP issue. :-)

The Tree / Server is only at startup - once we login a SLPINFO /D shows the Scope and DAs as up - as expected.

We have one site where setting Client Setting - IP Address Costing to 1 (subnet calc) made the issue go away. The default of 2 (ICMP ping) works for other sites - and the comms setup is the same !!! From this site we can ping the DAs and they respond within 100-300ms, well within 7.2 seconds timeout.

1. Anyone see any flaws in what is in place ?

2. Is doing a command line ping a valid test of the ICMP comms used by IP Costing method 2 ?

3. What the heck does the server setting "SLPDA Service Lifetime Window" do ? It mentions WAN connections so I am interested !

4. Would setting Servers to SLP TCP = ON help ?

5. Any other thoughts and tips !