SV: Re: Strange behaviour

[Chris Sharman]

>Both machines are Alpha's not clustered and the S_LOC include the NODE 
>and disk.
>I thought that with procloc you specify the default directory and you 
>could override that but it only works if procloc is on the same NODE 
>even if  i specify the full directory
>it dosen't work and if i include the NODE name i get the error 
>insufficient privilige or file 
>protection violation. I have PROXI in both machines can be that the 
>problem? it shouldn't

I don't know whether PH leaves it to RMS to sort out file locations, or does it
itself. I know in some instances it's a bit of both.

But RMS behaviour is this:

You have a filespec, a defaultspec, and/or optionally a 'related' spec (for
sticky defaulting).

Each spec has up to 6 components: node, device, directory, name, type, version.
The final file spec is the composition of the components found, so that
everything provided by the filespec is used; anything not provided is looked
for in the next spec.

So if, as I believe probable, the explicit filespec is used as the filespec,
and procloc is used as the default spec, then a node present in procloc
will always be used unless overridden by a node in the explicit filespec (ie
0::).

It's inherent in the design that once you turn a component on in a lower level
filespec, you can change it in a higher level, but can't turn it off.

Eg:
$ cre .test
Just a nameless .test
[exit]
$ copy .test test.test
$ copy *.test *.data
$ dir test,.test

The first spec "test" is parsed with default sys$disk:[]*.*;*, and finds
test.test and test.data.
The second spec ".test" is parsed with "test" as a related spec, so you don't
see the nameless files, just test.test again.

I don't know what performance you get out of this ? I'd be inclined to cluster
the machines myself. Failing that the suggestion of a procloc 'search list' is
a good one: define s_loc sys$login:,node::etc:[etc], or something. If the first
item in the search list is local you'll then get to search the local machine
first.

Chris
_______________________________________________________________________
Chris.Sharman@CCAgroup.co.uk		http://www.ccastat.demon.co.uk/
CCA Stationery Ltd, Eastway, Fulwood, Preston, Lancashire, PR2 9WS.
= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Subscribe: "subscribe powerh-l" in message body to majordomo@lists.swau.edu
Unsubscribe: "unsubscribe powerh-l" in message to majordomo@lists.swau.edu
This list is closed, thus to post to the list, you must be a subscriber.