Here comes a small update on the problem case after some testing yesterday:

-DNS, ICMP, Portstatus during Failover on both nodes > no problems, only 2 pakets lost

-Eventlogs during Failover > no errors

-Cluster Validation with Disks > no errors

-Connect via TCP and / or NP > no changes, Webapp spit our error 26 (ASP)

-change the dynamic TCP Ports for Named Instance from 60485 to 1433 > no change

-Test open UDP Port 1434 with portqry > check, ok

-check User Access rights for services and wepapp > ok

-SQL Browser restarts > no change

new in the situation is, that the databases in the default instance are all reachable, but with poor performance (on node 2). this only changes if i stop the sql brwoser on the passive node. also new: nearly all databases from the named instance are reachable now (including poor performance, sometimes), only one older web app cant reach their databases at all, no matter if tcp or np. this web app uses the native client, that can not be changed. error stays on 26 - SQL Network Interfaces, error: 26 - Error Locating Server/Instance Specified

The "split brain" scenario can be dropped, for that wasn´t in our mind. we want to run the instances either on node 1 OR node 2, besides with any other constellation there would be a problem with the quorum.

the question is: whats the difference between node 1 and node 2? they´re the modell, same build, same everything. the services are starting fine, there are no errors, theres no network performance issue to measure...

i´m really stuck with this problem

Die Frage ist also nun: Was ist an dem zweiten Knoten so anders als an dem ersten?