[Info-vax] strange SSH problem

[Alex Rubens]

On Sunday, May 17, 2015 at 10:10:44 PM UTC-4, David Froble wrote:
> Phillip Helbig (undress to reply) wrote:
> > In article <mjao84$28dt$1 at news.kjsl.com>,
> > helbig at asclothestro.multivax.de (Phillip Helbig (undress to reply))
> > writes: 
> > 
> >> Checking out TCPIP 5.7, I checked something I've never checked before,
> >> namely SSH from one node of the cluster to another.  SSH from the
> >> cluster to a remote host and vice-versa (coming in through the TCPIP
> >> cluster alias) works fine.  SCP as well.  (This is all I ever need SSH
> >> for so I haven't tested it within the cluster.) 
> >>
> >> The problem presumably has nothing to do with the upgrade since I get 
> >> the same behavious whatever two nodes are involved, upgraded or not, 
> >> whichever is client and which server etc.
> >>
> >> I do use a key for SSH from the cluster to a remote system, but 
> >> otherwise password authentication.  So, as long as SSH is running, 
> >> nothing to set up, right?
> >>
> >> SSH from one cluster node to another usually fails; after three attempts 
> >> at entering the password, it gives up.  However, in one of about 20 
> >> tries, it did work.  (No, the password isn't wrong and works fine with 
> >> telnet or SET HOST/LAT within the cluster.)
> > 
> > More information: Of 5 accounts I tried, 3 work and 2 don't.  So, it's 
> > not random, but still puzzling.  The 2 accounts which don't work with 
> > SSH work with telnet and SET HOST/LAT.
> > 
> > I get:
> > 
> > Status:                   %LOGIN-F-NOTVALID, user authorization failure
> > 
> 
> Check the Authorize records ...
> 
> Look for types of access ...

can you ssh localhost on each cluster member? What about the cluster alias from each cluster member? Where is TCPIP$SSH_HOME pointing to?