We have 99 Novell 4.11 servers patched to SP9.0 and using DS 6.13. DS time and replica synchronization reports no errors. All Windows 95 users can access all servers but OS/2 users cannot access some of those 99. The servers that appear to be affected are on a LAN segment EE01A202 but these servers can be accessed by both clients from any other LAN segment and only Windows 95 on the same segment. If the internal netware routers are reset on a server on that segment the problem goes away for a short time but the server that has its router reset isn't always the same one, it could be one of 30 on that segment.

Have you checked your LAN with a LANalyzer? A faulty network card or cable can cause a lot of trouble. If you tried: load dsrepair -> advanced -> View remote server ID list -> enter -> Verify all remote IDs on all replica server ?

Sorry for the delay. We have had telecoms connect an analyzer to both the server and workstations but as yet they have not identified anything useful. Both network cards in AMSSB1 and AMSSB2 have been replaced. A Cisco switch connecting 50% of the affected users has also been replaced. The original OS/2 client error 8805 can no longer be reproduced and we can always successfully login to these servers now but a login could take up to 1 hour. When logging in the clients produce the red banner and slowly line by line display the context, attached server etc. If a reset router command is issued on the offending server while they are logging in, the login will complete immeadiately but the server from which the command is issued could be one of 28 and is always different.