Tool like TreeSize which can reflect upon true file size when deduplication enabled?

[localzuk]

I'm looking for a way to get an idea of which folders are taking up space on our server, but since we have deduplication enabled, TreeSize doesn't work - it thinks all the files are 4kb in size (as the files are all technically junctions I believe).

 

Does anyone know of a tool which can show this data for us? It'd help me go through and get rid of bulky old stuff that isn't needed any more.

[AlexB]

I have no way to test this and is completely theory.

 

It may run quite a bit slower, but have you tried running Treesize against the folders over a network share? It may report differently (ie not junctions) when not scanning a local disk...

[sted]

I have no way to test this and is completely theory.

 

It may run quite a bit slower, but have you tried running Treesize against the folders over a network share? It may report differently (ie not junctions) when not scanning a local disk...

 

the free version at least wont run on network drives

[localzuk]

I have no way to test this and is completely theory.

 

It may run quite a bit slower, but have you tried running Treesize against the folders over a network share? It may report differently (ie not junctions) when not scanning a local disk...

 

It wouldn't make any difference. "Size on disk" always reports as 4k, regardless of where you look at a file, and its this size which Treesize uses.

[AlexB]

have you tried windirstat? (guess it depends which part of treesize you want)

[localzuk]

Yup, tried Windirstat.

[localzuk]

Hmm... Just realised I hadn't tried the latest version of Treesize Free.

 

It has the ability to list by "Size" rather than "Allocated".

 

It doesn't take into account deduplication (ie. if a file has 3 copies on disk, all 3 have a size) but its better than nothing.

[CamelMan]

Mapping a drive to a network folder and scan that. Works fine here.

[kafka-esk]

the free version at least wont run on network drives

 

On a side-note: The newest version of the freeware does, I think it's the same feature they use in the personal version (no Windows Servers, but network shares work). I was quite happy to see that.

Source: Scan Network Drives

Edited November 14, 2014 by kafka-esk

[AlexB]

Finally moved to a file server with dedup and now have the same issue.

 

Yes I can use treesize, but I'd rather not

 

Taking windirstat 1.1.2 source (last official release to my knowledge) I finally managed to get it to compile and then set about changing the filesize it reads. I ended up changing GetCompressedLength in FileFindWDS.cpp to this:

 

// Wrapper for file size retrieval
// This function tries to return compressed file size whenever possible.
// If the file is not compressed the uncompressed size is being returned.
ULONGLONG CFileFindWDS::GetCompressedLength() const
{
// ACB: Knacker this to only provide file size (not size of disk) for dedup checking
struct _stat64 buf;

if ((_stat64(GetFilePath(), &buf)) == 0)
{
	return buf.st_size;
}
else
{
	return 0;
}
}

 

For those not inclined to change the source and compile themselves, attached is my compiled version, no idea if it will work for you.

windirstatDD.7z

Edited August 24, 2015 by AlexB
changed version number to 1.1.2DD

[linkazoid]

Thanks, your version works great here!

[rpinto]

Many thanks for sharing ! It works fine

[GrimD]

Nice work! Just what I was looking for and works great!

[cradiotest2016]

Finally moved to a file server with dedup and now have the same issue.

 

Yes I can use treesize, but I'd rather not

 

Taking windirstat 1.1.2 source (last official release to my knowledge) I finally managed to get it to compile and then set about changing the filesize it reads. I ended up changing GetCompressedLength in FileFindWDS.cpp to this:

 

// Wrapper for file size retrieval
// This function tries to return compressed file size whenever possible.
// If the file is not compressed the uncompressed size is being returned.
ULONGLONG CFileFindWDS::GetCompressedLength() const
{
   // ACB: Knacker this to only provide file size (not size of disk) for dedup checking
   struct _stat64 buf;

   if ((_stat64(GetFilePath(), &buf)) == 0)
   {
       return buf.st_size;
   }
   else
   {
       return 0;
   }
}

 

For those not inclined to change the source and compile themselves, attached is my compiled version, no idea if it will work for you.

 

Come on guy, this is great!!!! Can you please add your patch to this bug issue in WinDirStat? https://bitbucket.org/windirstat/windirstat/issues/37/deduplication-support-for-windows-2012

[jmyers111]

Dude you are seriously my hero. This has been bugging me for years!

[crodgers]

To echo jmyers AlexB you are my hero - finally a way to properly report on file sizes on deduped volumes. Thanks!

[SiGmA_X]

Thank you, AlexB!

[Ant303]

i concur! bloody hero!!!! i almost spent $60 on folder sizes, but this is exactly what i wanted WinDirStat to do for me and now it does!

 

thank you so much!

[Norphy]

Why didn't I see this two and half years ago?!

[Delboy1978]

Hey gents,

 

I have downloaded this edited version of WinDirStat, what difference does this version do instead of the standard WinDirStat, it appears I am getting the same information back when the application is run?

 

Thanks

Del

[AlexB]

Hey gents,

 

I have downloaded this edited version of WinDirStat, what difference does this version do instead of the standard WinDirStat, it appears I am getting the same information back when the application is run?

 

Thanks

Del

 

The alternative version works if DeDuplication is enabled on the drive. There is no reason to use it unless you're using the dedupe built into Windows Server.

[AlexHeylin]

@AlexB - you're a star! Thanks :-D

[portwajn]

Yes I can use treesize, but I'd rather not

 

....

For those not inclined to change the source and compile themselves, attached is my compiled version, no idea if it will work for you.

 

AlexB, many thanks! Works great for me, Win2012R2 with dedupe.

 

FYI, I haven't found any comparable tool yet. However treesize actually shows correct sizes for deduped folders in right side pane (not in left pane actual tree - there it shows the wrong 'on disk' sizes). Unconvenient, but may also work for some, plus treesize seems to be much faster than windirstat.