Disk space allocation

[Tim Nelson]

I am working on some scripts to report on disk allocations and usage.
The show dev d /mount command gives me the currently free count and the output is great to work with, i.e. line by line space separated. The one additional item I need is total allocated to that disk / file system. I would love something that looked like a bdf or df -k output.
I apologize if I have missed the easy answer in the doc or help.
Thanks in advance.

[Ian Miller. · Accepted solution]

see F$GETDVI("devicename","FREEBLOCKS")
and F$GETDVI("devicename","MAXBLOCKS")

for example of use see
http://dcl.openvms.org/stories.php?story=04/01/11/4165701

[Tim Nelson]

THanks again Ian for the quick answer. I have not quite got a handle on some of the tricks with the functions.

If I could give you 100 pts I would.

Thanks !!!! Have a great weekend.

[Lokesh_2]

Hi,

See the online help for lexical f$getdvi . You can extract many more information about the disk.

$help lexical f$getdvi

HTH,
Best regards,
Lokesh Jain

[Tim Nelson]

Thanks Lokesh..

As you can tell I am an HPUX guy but not yet a VMS guy.

Thanks for the direction.

[Hein van den Heuvel]

Using the lexical f$getdvi in a DCL script is definetly the VMS style / native solution.
Btw... be sure to check out lexical f$device to get you a list of intereting (mounted!) deviced to loop through.

If you want to solve the problem more in a unix style then just use SHOW DEV/FULL/MOUN
and parse for "Total blocks 999999"
and "Free blocks 99999".

btw... I suspect that the 'real problem' you are trying to solve has been solved many times before, both in freeware and commercial packages. Still... this could be a fun excersize to get into VMS!

Enjoy,
Hein.

[Tim Nelson]

Thanks Hein, You hit it on the head. Cannot learn for myself if I use someone elses utility.

Thanks for the tips!!!

[Ian Miller.]

I guess a unix style answer is
pipe show dev/mount/fu | sear sys$input "Total blocks","Free blocks","Disk"

(or I guess this could be done in perl with the VMS::getdvi module)
but you don't want to do that do you :-)

Write some DCL using F$GETDVI - its fun and educational!

See http://dcl.openvms.org for examples.

[Tim Nelson]

Actually Ian. This is what I am working on.
Instead of stuggling the ins and outs with DCL scripting I am going to run some remote shells from a UNIX system to VMS so I can take the output and send it through awk so I can get some quick reporting done.
Someother day will be spent learning DCL.

Thanks !

[Ian Miller.]

awk is available for VMS
http://vms.process.com/scripts/fileserv/fileserv_search.exe?package=GAWK&description=&author=&system=Either&language=All&RD=&RM=&RY=

or, as I said you can do it in perl using the VMS::getdvi module

[Tim Nelson]

Even better.. See I am learning loads by the minute.

Thanks !!

[Ian Miller.]

and a pre-built perl can be obtained from
http://h71000.www7.hp.com/freeware/freeware60/perl-5_8_1/

Use the VMS::Device module (included in above)

[labadie_1]

Gawk comes with Vms (or more precisely with Tcpip 5.any)

$ gawk :== $ sys$common:gawk.exe_alpha

It is a strange idea to put it in the Snmp examples directory, but it is available.

[labadie_1]

Instead of perl, you can use python which has other qualities (readability, easy to maintain...)

python for Vms is available at

http://vmspython.dyndns.org/

see too
http://www.ferg.org/python_presentations/python_intro_for_managers.ppt
> http://www.ferg.org/python_slides/index.html

[Jan van den Ende]

I have posted 2 procedures at DCL.OPENVMS:

SPACE.COM is activated at a terminal(-session). It presents an overview of disk usage 1 line per disk.

MONITOR_DEVICES.COM can also be run at a terminal session, but it is INTENDED to be ran in a MOTIF display window. It presents disk usage as a percentage bar, but also checks clusterwide mount statusses & shadow statusses. Non-nominal situations are highlighted.
It also gives management a nice 'moving picture', so they can show 'everything is permanently under control' (which it actually DOES help with!).

Try if you like it, and DO let me know if you run into a configuration that is NOT covered.

Enjoy.

Jan

[Wim Van den Wyngaert]

Except monitoring free space, you should also check contiguous disk space. Programs may fail if space is too fragmented.

I don't know a method for retrieving this info in DCL so I try to create a contiguous file of 10.000 blocks every 30 minutes on almost every disk. If that fails : alarm.

[Willem Grooters]

Freewware product svailable on CD (5.0, perhaps 6.0 as well) doing this is FREE. One line per disk.
Possibly, DFU (freeware as well) may show the same issues (and defragments as well).

[Wim Van den Wyngaert]

Thanks for the tip Willem. It is only sad to hear that DCL can not do the job.

[Hein van den Heuvel]

Actually... specifically in a cluster, creating a large contiguous file as a test is probably the best method. (COPY/ALLOC=10000/CONT NL: dev:[000000]contig.tmp)

Cluster members have freeblock caches and the xqp will tell the other nodes to give back the free block cache before failing a request. [hmmm... not 100% sure it does this for lack of contig space]

Also... conceivably, if you fail to allocate the big file, then maybe you could release one of a few 'spare' contig space you pre-allocated and really soudn the alarm if you find your spares already released? ( $DELE dev:[000000]contig.tmp. )

groetjes,
Hein.

[Thomas Simpson]

Tim - Try looking in my directory for the file:
DISK_CAPACITY.COM
It produces a formatted report in GB. I have other variants if you really want it in blocks...

It also handles volume sets correctly.

Regards,
Tom

[Tim Nelson]

Thanks to all who responded