Here is the diskstat 1.0 package (157KB zip file).
A new tool has been ditributed, DISKSTAT. It will generate a report for the consequences of changing the clustersize of a disk. The program can also generate a filesize distribution list. See the freeware_readme.txt
I am very interested in your ideas about DISKSTAT. For example what you would like to be included in a next version, so please send your questions and/or feedback to diskstatdev@oooovms.dyndns.org.
Use it at your own risk, because we cannot take responsibility for crashes, dumps or even worse: loosing data.
010 31-may-2010
Initial version
Diskstat,utility, A program to display consequences of changing clustersize
DISKSTAT can do three things
1. Report of the consequences of a change to clustersize
2. Report on filesiz distribution
3. Report on specific file type
Author: Fekko Stubbe
Please mail problems/wishes to : dixdev (at) oooovms.dyndns.org
Ad 1.
This utility displays the consequences of changing the clustersize of a
disk. It will not change the clustersize, but only report.
If you enlarge the clustersizem, you will more blocks allocated, but unused
the gap between the EOF block and the Allocated size will increase.
IF you have a lot of small files, increasing the clustersize will have a
bigger impact than when you have only one very big file.
Diskstat needs read access to the indexf.sys on the disk.
DISKSTAT can also report on size distributions
Usage
Define a foreign command
$diskstat:=$device:[directory]diskstat
To report on clustersizes you can specify
$diskstat device /cluster[=(size1,size2,size3...)][/output=filename]
Default list is 1,2,3,4,8,16,18,32,35,64,70,128
Default output is sys$output
Example:
$diskstat/clus dsa0:
Cluster size 3 Total blocks 9086133 waste = 0.40% Current
Cluster size 1 Total blocks 9049753 waste = 0.00%
Cluster size 2 Total blocks 9075892 waste = 0.29%
Cluster size 3 Total blocks 9086133 waste = 0.40%
Cluster size 4 Total blocks 9126944 waste = 0.85%
Cluster size 8 Total blocks 9245400 waste = 2.16%
Cluster size 16 Total blocks 9495856 waste = 4.93%
Cluster size 18 Total blocks 9556866 waste = 5.60%
Cluster size 32 Total blocks 10062368 waste = 11.19%
Cluster size 35 Total blocks 10161095 waste = 12.28%
Cluster size 64 Total blocks 11290240 waste = 24.76%
Cluster size 70 Total blocks 11490710 waste = 26.97%
Cluster size 128 Total blocks 13848832 waste = 53.03%
The waste is calculated from a clusterize of 1 (hence the 0.0% on clustersize 1)
The first line is the current clustersize.
So if you would increase the clustersize for this disk to 16, you would
waste 4.93%(counting from clustersize 1) or 4.53%(from the currens clusterize)
ad 2.
Diskstat can also report on the filesize distribution
$diskstat device /size=(size_opt,size_opt...)[/output=file] [/allocated]
Diskstat will count filesizes in a table. Each table element has a
maxvalue. All files upto this maxvalue(and larger than the previous
tableelement) will be counted in this table element.
Normally DISKSTAT will use the EOF size (except for indexed files),
If you specify /ALLOCATED, DISKSTAT will use the allocated size.
Size_opt can be
tablesize=n The table has n elements, default is 8192
increment=n Each element is n blocks
factor=n Each element is n times as large as its predecessor
values=(v1,v2,v3) We begin with a list of v1,v2,v3 and then multiply
these values by the factor value.
All Report empty elements too
Examples:
increment=3 : Elements 3,6,9,12,15,....
Factor=2 : Elements 2,4,8,16,32,...
Values=(1,3,5,8),factor=10 : Elements 1,3,5,8,10,30,50,80,100,....
Default is increment=1
The output can be formatted in three ways
1. The default is lowval-highval:count (if increment=1 val:count)
and as as much as will fit on a line.
2. One table elements on one line (size_opt=oneline)
3. A csv type report, this might be used for Excell.
size_opt=csv[=separator]. Separator is
space comma tab colon semicolon bar slash or character=nn
Examples
$diskstat dsa0:/size : Will report all sizes per block
0: 471 1:6510 2:2559 3:5790 4:1564 5:1412 6:3051 7: 746
8: 641 9:1817 10: 512 11: 455 12:1078 13: 449 14: 374 15: 616
......
7957: 1 7958: 1 8039: 1 8157: 1 >8192:92(5293727 blocks,max=2073708)
Found 45385 files, total nblock 9049755, average size 199
$diskstat dsa0:/size=incre=10
0- 9:24561 10- 19: 5431 20- 29: 3036 30- 39: 2318
40- 49: 1274 50- 59: 885 60- 69: 1077 70- 79: 1376
......
52600-52609: 1 65530-65539: 1 77550-77559: 1
>81920:8(3677696 blocks,max=2073708)
$diskstat das0:/size=factor=3
0- 2: 9540 3- 8:13204 9- 26: 9455
27- 80: 7791 81- 242: 2887 243- 728: 1390
729- 2186: 702 2187- 6560: 303 6561- 19682: 77
19683- 59048: 26 59049- 177146: 7 177147- 531440: 1
531441- 1594322: 1 1594323- 4782968: 1
Found 45385 files, total nblock 9049757, average size 199
$diskstat dsa0:/size=(value=(1,3,5,8),factor=10)
0- 0: 471 1- 2:9069 3- 4:7354
5- 7:5209 8- 9:2458 10- 29:8467
30- 49:3592 50- 79:3338 80- 99: 621
100- 299:2636 300- 499: 681 500- 799: 440
.....
1000000- 2999999: 1
Found 45385 files, total nblock 9049757, average size 199
$diskstat das0:/size=(factor=3,csv)
0,2,9540 lowval,highval,count
3,8,13204
....
531441,1594322,1
1594323,4782968,1
$diskstat das0:/size=(factor=3,csv=char=64) !use the @ as separator
0@2@9540
3@8@13204
....
531441@1594322@1
1594323@4782968@1
Ad 3.
DISKSTAT can report on specific files
$DISKSTAT device/report=reporttype[/statistics]
/statistics will print counts of the report types
reporttype can be
overallocated File with eof<allocated-clustersize
organization=org Files with specific organization,
org can be: indexed,seqential,relative,other
format=fmt Files with specific record format
fmt can be: udf,fix,var,vfc,stream,stmlf,stmcr,other
mask=mask Report only files (wildcard-)matching "mask"
directories Report directories
notcontig Report non-contiguous files
empty Report empty files
mark_deleted Report file marked for delete
multiheader Report multi header files
full_name Report full filename instead of the name in the fileheader
This will result in more resource usage
Examples
$diskstat dsa0:/report=mark !report mark deleted files
Markdeleted DCLTABLES.EXE;830 Size 1436
Markdeleted DCLTABLES.EXE;765 Size 1437
$diskstat dsa0:/report=(mark,full) !the same but now with full filename
Markdeleted DISK$SYS_ITV002:[VMS$COMMON.SYSLIB]DCLTABLES.EXE;830 Size 1436
Markdeleted DISK$SYS_ITV002:[VMS$COMMON.SYSLIB]DCLTABLES.EXE;765 Size 1437
$diskstat dsa0:/report=mark/statis !add some statistics
Disk _DSA0: Volume SYS_ITV002 MAX_FILES = 1047510
Fileheaders (max_files ) total 1047510 used 45719 free 1001791( 4.36%)
Current size indexf.sys = 157410 blocks, clustersize 3
Fileheaders (Indexf size) total 157141 used 45719 free 111422( 29.09%)
Extentarea for indexf.sys, used 13 avail 155( 8.39%)
#blocks used 8843288 Blocks allocated 9164349 Waste = 3.63%
#Overallocated files 4541, Total size 78213
#Directories files 2787, Total size 2787
#Notcontig files 3110, Total size 2802165
#Markdeleted files 2, Total size 2873
#Multiheader files 26, Total size 2809259
#Empty files 471, Total size 0
Counts for file organizations
#Sequential files 45218, Total size 8984014
#Relative files 2, Total size 18
#Indexed files 160, Total size 65718
#Other files 5, Total size 7
Counts for record types
#Undefined files 39, Total size 1349249
#Fixed files 4597, Total size 5485927
#Variable files 10989, Total size 602300
#VFC files 1088, Total size 38590
#STream files 2867, Total size 110960
#StreamLF files 25803, Total size 1462720
#StreamCR files 2, Total size 11
#Other files 0, Total size 0