Re: [Exim] Re: spool over NFS?

Startseite
Nachricht löschen
Nachricht beantworten
Autor: Peter D. Gray
Datum:  
To: exim-users
Betreff: Re: [Exim] Re: spool over NFS?
I think most of the information I have seen in this thread is
correct, but the circumstances for it being correct have
not been stated.

From my knowledge:

- if you use NFS version 3 AND you set the timeout
    on the attribute cache to 0 (no attribute caching)
    then there is no problem using NFS for anything
    as long as you do your locking properly.


    This is because NFSv3 has an atomic file create
    operation which allows .lock files to work
    ala UFS. Also, attribute information is returned
    with every aknowledgement.


    But, if you fail to get a fcntl lock, you must
    close and re-open the file before trying to get the
    lock again.


- if you use NFSv2 the .lock files are hard to make work
    (lack of atomic create).
    Philips comments on this are probably correct.
    You can use a directory for locking because
    NFSv3 does have an atomic mkdir operation.


- if your implementation of NFS on either client or
    server is broken, then obviously all bets
    are off. For a long time many clients and servers
    had problems with fcntl locks, but I kind
    of hope everybody has this working by now.


- if you use NFSv4 then the semantics of NFS are pretty much
    identical to local disk (I think). You do not need
    to do anything special (like play with the
    attribute cache timeouts) to make things
    work properly.


Does anybody with superior knowledge want to correct
any of the above?

Regards,
pdg

--

See mail headers for contact information.