topic Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value? in Operating System - Linux

Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Alzhy — Tue, 07 Feb 2006 12:18:21 GMT

I'm dealing with hundreds/thousands of cXtYdZ values on a typical server and I have a need in my script to pick from a list of cXtYdZ the lowest numbered cX, then the lowest numbers tY then the lowest numbered dZ -- in effect the lowest numbered/valued cXtYdZ.

Suggestions? More points to Perl constructs.

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Steven E. Protter — Tue, 07 Feb 2006 12:21:20 GMT

Shalom Nelson,

Perl from someone else

ls | /dev/dsk/c* | sort -rn

take out the r if it comes out backwards.

SEP

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Rick Garland — Tue, 07 Feb 2006 12:30:57 GMT

perl or no perl, you still run it through the 'sort' command

In perl, make a call to a system command

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Bharat Katkar — Tue, 07 Feb 2006 12:32:46 GMT

ioscan -funC disk | grep c | grep -v Class | awk '{ print }'

Does this output not sorted.
If i understood the question the output of the command above is what you are expecting.

Regards

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Bharat Katkar — Tue, 07 Feb 2006 12:34:45 GMT

sorry typo. it's like this:
# ioscan -funC disk | grep c | grep -v Class | awk '{ print $1}'

Regards

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Hein van den Heuvel — Tue, 07 Feb 2006 12:58:32 GMT

The problem with a straighy sort is that for example t10d sorts before t2d.
So you need to pull apart the X, Y and Z.

Now you can create a perl sort function that first compares X then Ys, then Zs, or you can cheat a little.
Create a fresh number by multiplying X by a large range (like 1000... but take your pick), then add Y and mutliply again, and finally add Z.
Now compare numerically.
Pick any number larger than the largest X and Y, but pikcing 1000 or 100 reads nicely.

Sample script to get you going:

------ sort_dsk.pl -----
while (<>) {
if (/dsk.c(\d+)t(\d+)d(\d+)\s/) {
$num = ($1*1000 +$2)*1000+$3;
$dsk{$num} = $_;
}
}
for $key (sort {$a <=> $b} keys %dsk) {
print $key, $dsk{$key};
}
-------------

As input example I used the data from the last reply in: http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=718178

This then gives:

$ perl sort_dsk.pl tmp.tmp
3000000/dev/rdsk/c3t0d0 0 0 CL1F 1:71 OPEN-3 2347 00030107 5247
3000001/dev/rdsk/c3t0d1 0 1 CL1F 1:72 OPEN-3 2347 00030107 5247
3000002/dev/rdsk/c3t0d2 0 2 CL1F 1:73 OPEN-3 2347 00030107 5247

:
6004001/dev/rdsk/c6t4d1 4 1 CL1F 2:7e OPEN-9 7042 00035807 5244
6004002/dev/rdsk/c6t4d2 4 2 CL1F 2:7f OPEN-9 7042 00035807 5244
6009000/dev/rdsk/c6t9d0 9 0 CL1F 3:2c OPEN-9 0 00035807 5244
6010000/dev/rdsk/c6t10d0 10 0 CL1F 2:2e OPEN-9 0 00035807 5244
6011000/dev/rdsk/c6t11d0 11 0 CL1F 1:80 OPEN-9 0 00035807 5244
6012000/dev/rdsk/c6t12d0 12 0 CL1F 1:88 OPEN-9 0 00035807 5244
6013000/dev/rdsk/c6t13d0 13 0 CL1F 3:74 OPEN-9 0 00035807 5244

Enjoy!

Hein.

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

James R. Ferguson — Tue, 07 Feb 2006 13:13:18 GMT

Nelson:

# perl -nle 'm/c(\d+)t(\d+)d(\d+)/;printf "c%02dt%02dd%02d\n",$1,$2,$3' file|sort

Regards!

...JRF...

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Alzhy — Tue, 07 Feb 2006 13:30:04 GMT

# perl -nle 'm/c(\d+)t(\d+)d(\d+)/;printf "c%02dt%02dd%02d\n",$1,$2,$3' list|sort
c01t03d00
c111t03d00
c20t03d00
c211t03d00
c21t03d00

CLose but it adds leading zeroes...

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Alzhy — Tue, 07 Feb 2006 13:32:40 GMT

The result I'd like from the test list should be:

c1t3d0
c111t3d0
c20t3d00
c21t3d0
c211t3d0

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

James R. Ferguson — Tue, 07 Feb 2006 13:51:03 GMT

Hi (again) Nelson:

Of course:

# cat listdevs
#!/usr/bin/perl
while (<>) {
m/c(\d+)t(\d+)d(\d+)/;
$dev = sprintf "c%04dt%04dd%04d\n",$1,$2,$3;
push @ary, $dev;
}
@sorted = sort @ary;
foreach (@sorted) {
m/c(\d+)t(\d+)d(\d+)/;
printf "c%dt%dd%d\n",$1,$2,$3;
}

...run as ./listdevs file

Regards!

...JRF...

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Hein van den Heuvel — Tue, 07 Feb 2006 14:02:41 GMT

Didn't my example give you enough hints to get going?

If you just want the device list then here is a minor adaption to that...

--------------
while (<>) {
if (/c(\d+)t(\d+)d(\d+)\s/) {
$num = ($1*1000 +$2)*1000+$3;
$dsk{$num} = "c$1t$2d$3";
}
}
for $key (sort {$a <=> $b} keys %dsk) {
print $dsk{$key}."\n";
}
--------------------

fyi... The {$a <=> $} is a localy defined sort function using a numeric compare instead of the default string compare.
Due to the way i contractured the artifical 'key', either should work.

------------
while (<>) {
if (/c(\d+)t(\d+)d(\d+)\s/) {
$num = ($1*1000 +$2)*1000+$3;
$dsk{$num} = "c$1t$2d$3";
}
}
for $key (sort keys %dsk) {
print $dsk{$key}."\n";
}

Re: Best and fast way to sort cXtYdZ values picking the lowest "c" value?

Andy Torres — Wed, 08 Feb 2006 08:28:41 GMT

Here's a one-liner, but you'd have to play with the "sort" part to accomodate single & double-digit c#'s:

ioscan -funC disk | grep dsk | awk -F / '{print $4}' | sort | head -1