Max Files Question

[jeffc]

Is there a default limit to the maximum number of
files that can be created in a directory under
OpenVMS 7.1?

Would this be something controlled when the directory
is created or when the actual disk is initialized?

Thanks,

<Jeff Cooper -- je...@iglou.com>

[avaj...@gtech.com]

In article <3555AFFC...@iglou.com>,

je...@iglou.com wrote:
> Is there a default limit to the maximum number of
> files that can be created in a directory under
> OpenVMS 7.1?

No.

But, there are limits for how many file ther can be on a disk
based on phsyical disk size and INDEXf:SYS limits.

Also note that large-directories slow down access significantly
because directories over a certain size is not cached !

Arne

-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading

[Gotfryd Smolik, VMS lists]

On Sun, 10 May 1998, jeffc wrote:

+Is there a default limit to the maximum number of
+files that can be created in a directory under
+OpenVMS 7.1?

AFAIK no !
But you may know of the 128 block size limit (bigger directories
are not cached !), of the "non-span" (there are a "holes"
in the file) and "continuos" requirements. You must have
place (continuos !) for the directory.
Anyway, summarising all discuision here: creating a directory
with big number of files isn't (under VMS) good idea !

+Would this be something controlled when the directory
+is created or when the actual disk is initialized?

On the init time you control the general limit of files
on the whole volume, and this is differrent point...
Anyway you can have *more* files in directory than limit
of files on volume (MOUNT/BIND !! -;>).

On *some* AlphaVMS have seen that can be used:
$ CRE/DIRE [.subdirname]/ALLOC=size
where have work (check on VAX VMS 6.2: don't find)
and can be usefull if you must get thru the worsest way...

+Thanks,
+
+<Jeff Cooper -- je...@iglou.com>

Regards - Gotfryd

--
=====================================================================
$ ON F$ERROR("LANGUAGE","ENGLISH","IN_MESSAGE").GT.F$ERROR("NORMAL") -
THEN EXCUSE/OBJECT=ME
$! G...@stanpol.zabrze.pl
=====================================================================

[Hein RMS van den Heuvel]

In article <3555AFFC...@iglou.com>, jeffc <je...@iglou.com> writes...

>Is there a default limit to the maximum number of

>files that can be created in a directory under OpenVMS 7.1?

No formal limit really, it is more what performance level you can stand.

The only hard limit that I know of is that the size of a directory
is limited to 64k blocks. If you have short file names, and many
versions of the same file name, then you could get about 60 files per
directory block. I think this comes to something over 3 million files.
Eventually (and long before 64k blocks) you will probably not like
the performance. 3 million files means 3 million file header
blocks (at least), so you would have 1.5 gigabytes of your disk
occupied even if the files were zero length - with data in them you
should multiply that by a bit more than the disk cluster factor to find
out how large a disk you would need.

There was a (long since fixed in VMS V5.2ish) bug in VMS that caused
'service disruptions' at 1800 blocks. Only few noticed that in those days.

Depending on the application access patterns, you may find a noticeable
performance degradation as you cross 128 blocks. If you expect that to
be the case, then you may want to design with subdirectories from the get go.

hth,
Hein.

[Jan Vorbrueggen]

vanden...@eps.enet.dec.c*m (Hein RMS van den Heuvel) writes:

> There was a (long since fixed in VMS V5.2ish) bug in VMS that caused
> 'service disruptions' at 1800 blocks. Only few noticed that in those days.

Ahh...I hadn't seen that euphemism before...must remember it...almost as good
as Boeing's "variance at completion" for "cost overrun"...

Jan

[Shane F. Smith]

--Boundary_(ID_Zd5BRxdfIH8Y9brytHoZ0A)
Content-type: TEXT/PLAIN

Or the BBC Micro repairman back in 84 who described my friend's machine as
"operating in a mode of non-functionality". B^)

##### ---------------------------------------------------------------
#-O-O-# | shane dot smith at earthling dot net, spam to ab...@127.0.0.1 |
# L # ---------------------------------------------------------------
#===# Don't blame HealthNet for anything I say. They're innocent.
### Bob Palmer for C.E.O of Microsoft!

--Boundary_(ID_Zd5BRxdfIH8Y9brytHoZ0A)--