On Wed, Mar 7, 2012 at 11:53 PM, Julian Simioni
<[email protected]> wrote:
> On Wed, Mar 7, 2012 at 6:57 PM, Frank Steinmetzger <[email protected]> wrote:
>>
>> Hello list
>>
>> It came to my attention that during (after) an emerge run, df reports
>> considerably less space available on my / than before the emerge (everything
>> except /home sits on the root partition). I was wondering how this comes to
>> be, since I have /var/tmp/portage on tmpfs.
>>
>> I am in the middle of a KDE upgrade (4.8.0→4.8.1) right now and before I
>> started, I downloaded all distfiles and then looked at df /, it showed 1022
>> blocks, hence about 1 GB of free disk space. I am at package 115 out of 174
>> right now, and df shows a mere 389k blocks remaining.
>>
>> Also before I began the emerge run, I started 'ncdu -x /' which scans all
>> dirs
>> on the / partition and then I can browse through my FS hieararchy, showing
>> the
>> disk usage of every directory. Now I ran the same ncdu command again in
>> another screen, so I can compare it with the first one.
>>
>> The folders themselves have 0.1 to 0.2 GB difference between their old and
>> new
>> state, and ncdu's bottom bar even shows the same values for both apparent and
>> real total disk usage (rounded to 0.1 GB). So what am I missing here? I
>> searched df's man page for something about apparent sizes/sparse files, but
>> then again, why would portage create such files in the first place?
>>
>> Do you have any thoughts that might help me understand what I'm seeing?
>> --
>> Gruß | Greetings | Qapla'
>> I forbid any use of my email addresses with Facebook services.
>>
>> You will find everything in an online database.
>> Just not what you are looking for.
>
> Unless you have it mounted on tmpfs for increased compilation speed as
> many others do, /var/tmp/portage can easily grow to several hundred
> megabytes as packages are compiled. Once the compilation finishes
> successfully, it will be cleaned up, so the contents are constantly
> changing during an emerge, and it may not be easy to track down after
> the fact.
And only after hitting send to I register the line where you mention
that you do in fact use tmpfs. doh!
