Conference smurf::ase

    Sid,
    
    What is the complaint of the customer?
    
    	- the fact that the asedirector is not killed ?
    	- the fact that he can call asemgr ans see the system status?
    	- the fact that both daemon logs shows a NET PARTITION ?
    	- the fact that in this NET PARTITION situation, he cannot 
    	  relocate, or restart any services ?
    
    The last point is the standard behaviour of ASE since the beginning.
    
    The first and second points may have change since the first version,
    but are not and objective per se, but a way to achieve the last point.
    
    The NET PARTITION seems obvious as the NET cable has been removed.
    
    Manu.

Manu,
        
    What is the complaint of the customer?
    
      If the director is supposed to exit when there is a net
      partition and it is not doing so - and if the CSC is insisting
      and ASE Engineering is insisting that this behavior is wrong
      then the customer has a right to be somewhat uncomfortabe
      about what is going to happen in a real life situation. 
      
      These people are running mission critical applications with
      DECsafe; they have a strong need to know what might happen when
      various failure events occur. If they are observing a different
      behavior than what Digital is telling them they should observe
      under these conditions then we would seem to owe them either an
      explanation or a patch.  They would be idiots to simply shrug
      off a discrepancy like this and hope that all works out when
      a real failure happens. They are running the company heart
      and soul and other sensitive bits on these platforms. Then
      need to know how they are and are not covered.
      
      d

    Hi Dave,
    
    Ok, I now  understand your concern,... and I am not able to confirm or
    infirm that DECsafe has changed versus the explnation given in the
    Admin guide. If you really need an explanation, I think you should open
    a QAR on this issue giving the ASE version and the ADmin guide version
    you refer to.
    
    I believed, at first, that you wondered what is the expected behaviour
    of DECsafe in case of NET partition, and thus I concentrated my answer
    on that. By the way, the behaviour of DECsafe is still the same and
    only the way to achieve it has changed. Maybe Greg, Mitch or Doug can
    shed some light on the version evolution... ?
    
    Regards, Manu.
    

re:   <<< Note 1928.3 by BRSDVP::DEVOS "Manu Devos DEC/SI Brussels 856-7539" >>>
                                   -< ... ? >-

   By the way, the behaviour of DECsafe is still the same and
    only the way to achieve it has changed.
    
    
    The externally observable behavior appears to be similar to the
    way it is 'supposed to be'. "Appears to be" is a bit shy of where
    one would like to be when dealing with a major corporation's
    sensitive parts.
    
    d

    Hi,,
    
    By re-reading notes 1907 and 1928 attentively, you are running UNIX
    3.2G and thus ASE1.3. This ASE 1.3 version has introduced the support
    of multi-networks for the inter-cluster communication (HSM_PATH_STATUS)
    and the network monitoring feature (HSM_NI_STATUS). HSM_PATH_STATUS is
    covering the network partition. So, it is most likely possible that the
    code has changedin this version 1.3 versus before 1.3.
    
    I have not the ADMIN guide of the version V1.2, only the ADMIN guide of
    version 1.3, so I cannot compare the difference. But, from my mind,
    before version V1.3, the asedirector was simply exiting in case of net
    partition. In the ASE 1.3 ADMIN guide AA-QP0TA-TE v1.3 on page 1-12, I
    can read:
    
    
    "For example, a full network partitionoccurs if only one network is
    used in an ASE and that path failes, or if more than one network is
    used and all the paths fail. If a full network partition occurs, the
    services continue to run on the member system and can be automatically
    failed over IF THE SYSTEM CRASHES, but you cannot use asemgr utility to
    change the ASE or to manually relocate services."
    
    In the appendix B1 and B2, I cannot see that the asedirector is exiting
    when a network partition occurs.
    
    So, according to the documentation of ASE 1.3, the asedirector is not
    exiting. If you have the chance to look in the ADMIN guide of ASE 1.2,
    then maybe you will find that the asedirector was killed in case of
    network partition. But, I hope that you agree that the BEHAVIOUR is not
    "appearing" to look the same, but is mentionned clearly in the ADMIN
    guide of ASE 1.3
    
    regards, Manu. 

re: .5

thanks.  I had read 1-12 but had not exactally inteperted it's full meaning
and the difference between 1.2 and 1.3   

Sid Johnson
Customer Support Center/Atlanta
[Posted by WWW Notes gateway]

re: .5
    
    
    Sounds good to me... It'd feel better if somebody from engineering
    would confirm it.