 |
The Question is:
I have always beleived that OpenVMS PID's are unique for as long as the system
is booted and are only re-used after a reboot. Is this true? And if so, is
there any limit to PID's - i.e. the 8 hex digits normally displayed? What
would happen if this lim
it were reached?
I know 8 billion unique processes should be enough for anyone and we are not in
imminent danger of hitting this limit, but I would still be interested in the
'what if' scenario, particularly as a few UNIX-types are siting this as a
weakness with OpenVMS
The Answer is :
With OpenVMS, the PID format (both external and internal PID varieties
in the current model) formats are undocumented and subject to change
without notice. The current format of the external and internal PIDs
are listed in the OpenVMS Internals and Data Structures Manual (IDSM),
though the OpenVMS Wizard strongly discourages any dependencies on PID
format be implemented in any application software.
As for what happens when the PID itself wraps, well, nothing particularly
interesting happens.
The value wraps.
Continous OpenVMS system uptimes of fifteen or more years are known to
have occured on customer systems, so there is some experience with very
extended system uptimes.
With the wrapping of the PID, you will have the potential for duplicate
PIDs in accounting and auditing files, particularly if you have not
recreated new versions of the accounting and auditing logs during the
intervening years. That said, due to the construction of the PID, you
cannot and will not have duplicate PIDs active on the same system or
same cluster at the same time. Also due to the construction of the
PID, a very long time will transpire between any particular uses of any
particular PID value.
UNIX has a very similar process identification (PID) construct.
|
|
|
|
|
