lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:	Tue, 22 Sep 2009 17:31:06 +0200
From:	jehan procaccia <jehan.procaccia@...sudparis.eu>
To:	Eric Sandeen <sandeen@...hat.com>
CC:	Theodore Tso <tytso@....edu>, linux-ext4@...r.kernel.org
Subject: Re: howto downgrade ext4 to ext3

Eric Sandeen a écrit :
> jehan procaccia wrote:
>> Other inodes incriminated showed the same result -> over-quota . So 
>> if user data finally cannot be written, after all ... quota wouldn't 
>> allow it anyway .
>
> The unfortunate thing in your situation though is that the user 
> -thought- it did get written, just FYI.
>
> When quota is working properly, the denial is obvious up front.
OK, indeed it's not fair for the users !
I had the time to experiment an ext4 FS mounted with -nodelalloc, and 
now quotas seems to work fine with it ;-) .

here's a test on a FS  mounted with -nodelalloc
while writing with dd for an amount of data between interval soft 300M 
and hard  330M, it worked fine:

[test@...-05 ~]$ dd if=/dev/zero of=./big-101M.dd bs=1024 count=101000
103424000 octets (103 MB) copiés, 9,06256 s, 11,4 MB/s
$ quota -s test
/dev/mapper/VolGroup03S2IA-LogVolS2ia01
                  317M*   293M    323M   7days    1760       0       0

Then, user "test" writes another 100 Mb and fails as expected :

[test@...-05 ~]$ dd if=/dev/zero of=./big-102M.dd bs=1024 count=102000
dd: fermeture du fichier de sortie `./big-102M.dd': Débordement du 
quota d'espace disque
$ quota -s test
/dev/mapper/VolGroup03S2IA-LogVolS2ia01
                  323M*   293M    323M   6days    1761       0       0
[root@...eh /disk00/test/test]
$ ll big*
-rw-r--r-- 1 test mci 103424000 Sep 21 11:11 big-101M.dd
-rw-r--r-- 1 test mci  35729408 Sep 21 11:13 big-102M.dd

however in terms of performances, it seriously drops (nearly 50% !)  
with -nodelalloc on the NFS disk server, here's a simple local dd write 
test:

ext4 -nodelalloc
[root@...eh /disk00]
$ dd if=/dev/zero of=./big-300M.dd bs=1024 count=300000
307200000 bytes (307 MB) copied, 1.20125 seconds, 256 MB/s

ext4 -delalloc
[root@...eh /disk01]
$ dd if=/dev/zero of=./big-300M.dd bs=1024 count=300000
307200000 bytes (307 MB) copied, 0.752382 seconds, 408 MB/s

however, the same test from an NFS workstation shows very small 
differences between partition mounted with nodelalloc and delalloc:

-nodealloc + NFS
$ dd if=/dev/zero of=./big-100M.dd bs=1024 count=100000
102400000 octets (102 MB) copiés, 9,09703 s, 11,3 MB/s

delalloc + NFS
$ dd if=/dev/zero of=./big-100M.dd bs=1024 count=100000
102400000 octets (102 MB) copiés, 9,00392 s, 11,4 MB/s

I supose that NFS is the bottleneck !

I will announce a maintenance on that server in order to remount  my 12 
x 200GB partitions with -nodelalloc, if ever performances drops too much 
, then I'll probably choose to set quotas off until a redhat kernel 
package provides a solution .

does 2.6.31 already patche to deal correclty with delalloc and quotas ?

Thanks .

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists