On 05/27/2012 08:09 PM, Henrique de Moraes Holschuh wrote:
> But my point was that people will not switch to writing small temporary
> files somewhere else, because they have not switched to writing files to
> $TMPDIR even after 10 years.
>
Please define who are the people you are talking about. If they aren't
using mktemp, please file a bug. Also, if we really have a critical
performance issue with /tmp, then having an alternative in /run/tmp
using tmpfs is the solution, and it really can (and will) be
addressed if there's a real problem.

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC22D4B.8080909@debian.org">http://lists.debian.org/4FC22D4B.8080909@debian.org

On 05/25/2012 05:41 PM, Josselin Mouette wrote:
> Le vendredi 25 mai 2012 à 11:04 +0200, Salvo Tomaselli a écrit :
>
>> Apparently not everybody has experienced that, and this would explain why they
>> are so happy about the idea of paging out a couple of Gigabytes.
>>
> Because paging out a couple Gigabytes is veeeeeery different from
> writing a couple Gigabytes to disk, of course.
>
Of course it is, and that's the whole point !!!

When a system is swapping, you can click, nothing happens...
If it's a server, that's even worse. You type "top" to see what
the fuck is going on, and then you wait ... Even in some cases,
you can't even log in through ssh (or it's so slow...).

If an application is blocked writing on /tmp, and you think it's
taking too much time and you want to kill it, open a terminal,
and type "xkill" (or just kill with the kill / killall command if
you're on a server). That's instant, and your system continues
to be responsive.

>From my experience, all of the above is also the case on a
relatively fast RAID10 (eg: 200 MB/s) with 4 cores...

IMO, if you didn't get that point, then you didn't understand
this thread at all.

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC2312A.9080307@debian.org">http://lists.debian.org/4FC2312A.9080307@debian.org

On 05/25/2012 03:29 PM, Josselin Mouette wrote:
> Which means a *huge* performance
> improvement. Do the measurements yourself, it works with basically
> anything that makes heavy use of /tmp.
>
I have yet to know what application you are talking about.
Who's doing heavy use of /tmp exactly?

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC231F4.60102@debian.org">http://lists.debian.org/4FC231F4.60102@debian.org

On 27/05/12 14:53, Thomas Goirand wrote:
> On 05/25/2012 03:29 PM, Josselin Mouette wrote:
>> Which means a *huge* performance
>> improvement. Do the measurements yourself, it works with basically
>> anything that makes heavy use of /tmp.
>>
> I have yet to know what application you are talking about.
> Who's doing heavy use of /tmp exactly?
>

$ du -s --si /tmp/temp-lintian-lab-7B9v6Qamm8/
448M /tmp/temp-lintian-lab-7B9v6Qamm8/

--
Regards,
Dmitrijs.


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC23303.60301@debian.org">http://lists.debian.org/4FC23303.60301@debian.org

On 05/25/2012 07:44 PM, Roger Leigh wrote:
> However, the majority of
> software which finds the tmpfs too small has unreasonable expectations
> of what can be expected to be available (by default).
>
We welcome you to rewrite / patch these software then!
Good luck, the list is huge, and growing every day this
thread is alive.

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC23E43.3020201@debian.org">http://lists.debian.org/4FC23E43.3020201@debian.org

On 05/26/2012 03:12 AM, Nikolaus Rath wrote:
> Aren't quotas only for non-root and per file system? I think we're
> already safe from non-root filling up / because of the reserved 5%.
>
> Best,
>
> -Nikolaus
>
That's a bit off-topic but...
XFS has project quota, which are directory based. I'd love to see this
kind of feature in extXfs.

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC2402A.4060406@debian.org">http://lists.debian.org/4FC2402A.4060406@debian.org

> Please define who are the people you are talking about. If they aren't
> using mktemp, please file a bug.

man 3 mktemp
BUGS
Never use mktemp() [...]


--
Salvo Tomaselli


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 201205272008.30649.tiposchi@tiscali.it">http://lists.debian.org/201205272008.30649.tiposchi@tiscali.it

Salvo Tomaselli dixit:

>> Please define who are the people you are talking about. If they aren't
>> using mktemp, please file a bug.
>
>man 3 mktemp
>BUGS
> Never use mktemp() [...]

Your manpage search order is (also) wrong:
‣ https://www.mirbsd.org/man.cgi?mktemp

(1) come before (3) section manpages. (And in this context, the entire
mkstemp(3), mkdtemp(3), etc. family is meant.)

bye,
//mirabilos
--
13:37⎜«Natureshadow» Deep inside, I hate mirabilos. I mean, he's a good
guy. But he's always right! In every fsckin' situation, he's right. Even
with his deeply perverted taste in software and borked ambition towards
broken OSes - in the end, he's damn right about it ! […] works in mksh


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: Pine.BSM.4.64L.1205271921230.24443@herc.mirbsd.org ">http://lists.debian.org/Pine.BSM.4.64L.1205271921230.24443@herc.mirbsd.org

On 05/28/2012 02:08 AM, Salvo Tomaselli wrote:
> man 3 mktemp
>
I was talking about /bin/mktemp.

Thomas


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 4FC28113.2070308@debian.org">http://lists.debian.org/4FC28113.2070308@debian.org

]] Thomas Goirand

> Come on, WE CAN! Let's create a /run/tmp *now*, it wont cost us much,
> then later applications can slowly use it if they want. Worst case:
> not a lot of app uses it, and the /run/tmp tmpfs isn't used much,
> which isn't a problem. Best case: everyone that needs a fast space
> has it and switches to it, and everyone is happy.

How is this significantly different from /run/user/$user ? If it's not,
we should just use that instead.

--
Tollef Fog Heen
UNIX is user friendly, it's just picky about who its friends are


--
To UNSUBSCRIBE, email to debian-devel-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Archive: 87k3zxwfzh.fsf@xoog.err.no">http://lists.debian.org/87k3zxwfzh.fsf@xoog.err.no