Hi Tom,

>I am investigating the possibility of running two versions of PMDF
>within s single cluster. The reason is to ease in a new version in a
>rolling upgrade rather than change everything at the same time.
>...
>The main concern is to do with communication across the cluster, mainly
>locking stuff. Is there anything there that would mean say a 6.2-1 and
>6.3 (or 6.3 & 6.4) PMDF machine should not be clustered with a common
>set of queue files ?


I have my suspicions that the message counter process will get into trouble,
based on something I noticed here when I ran PMDF 6.3 on one system and PMDF
6.4 (beta) on another.

Regards,

Jeremy Begg