[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

EXT3, REISERFS: Beberapa tambahan info



Bapak-bapak dan Ibu-Ibu;

Mungkin ingin berbagi pengalaman menggunakan reiserfs
atau ext3 untuk sistem produksi? Saya sendiri pake keduanya
untuk keperluan pribadi/beban ringan tanpa masalah apa-apa.
Nah, seberapa aman untuk sistem produksi heavy-duty?

Terlampir tambahan (baru) informasi yang dicomot dari 
milis sana-sini.

--
Abdul Latip -- Angkasa Internet Junior Staff -- ANGIN.com
http://people.WebIndonesia.com/dullatip/ ----------------

=============================================
(Klaus Imgrund Tue, 20 Aug 2002, debian-user)

Well,
had a problem for some time when my machine was going overnight.
In the morning there was either a black screen where there used 
to be a desktop or a blank screen - could get anywhere any more.
Today it finally quit on me - filesystem corrupted.
My partitions go like this:

/dev/hda1   fat
/dev/hda2   ext3  debian
/dev/hda3         swap
/dev/hda5   ext2  data
/dev/hda6   ext2  gentoo

All partitions where mounted and only hda2 got corrupted - fixed 
it from gentoo. The hdd is about 6 month old and doesn't give me 
any trouble at all under normal conditions. According to the logs 
the machine was doing nothing - no errormsg's. The powersupply is 
ok. I did have some strange lines in the root-terminal - no idea 
where those came from. Since the install is pretty much hosed I 
am wondering what to do next. Go with ext3 again or try something 
else? I'd appreciate it very much if anybody could give me an 
idea what is going on here.


=========
(patrick)
I had the same thing happen recently, random file corruption on 
an ext3 partition then lost the entire partition table.  

I'm pretty sure the problem was with the hard drive tho, not 
ext3 as I had a second, identical IBM hd fail completely at the 
same time.  (Third drive, a western digital was fine)

FWIW, a friend who sells systems tells me he's seeing an awful 
lot of bad Fujitsu drives right now...


===============
(Klaus Imgrund)
Ok - but the thing is that all partitions where mounted and 
only /dev/hda2 got hosed. There aren't a lot of things working 
anymore there.The other partitions are just fine - not one file 
corrupted there.  Thats what I don't understand.


==================
(Jamin W. Collins)
Odd, I've been using ext3 on all my systems since ~2.4.7 
and haven't had any file corruption that I can in any way 
attribute to ext3.  No way I'd go back to plain ext2.


===============
(Klaus Imgrund)
Well no more ext3 for me either.
Is reisser supported with woody?
I did some looking around and didn't find any info on the subject,plus I
can't reach debian.org right now.


============
(Mike Kuhar)
Hmmm, interesting.  I had the same problem with an ext3 partition 
on a Maxtor drive.  Got to the point where I had to re-install.  
I was using kernel version 2.4.17 at the time.  I think the 
problem is with ext3. Since I re-installed and went back to ext2 
on the same drives, I've had no further problems. I've also gone 
through kernel versions 2.4.18 and currently on 2.4.19.  I might 
add, however, I have had ext3 on my laptop since 2.4.17 with no 
problems.  Besides different drives, there is only one partition  
on the laptop, whereas my desktop has 4.(not including swap).
I don't know about you, but I'm steering clear of ext3 until
some time in kernel 2.6.x,  -mk


==============
(Eduard Bloch)
Reiser's journalling mode is even less safe and will corrupt data 
faster on broken hardware.

> Is reisser supported with woody?
Yes, with the bf2.4 installation flavor.

> I did some looking around and didn't find any info on the subject,
> plus I can't reach debian.org right now.
Network problems, just wait.


============================================
(R. Bradley Tilley Debian User, 24 Aug 2002)

We're using reiserfs on debian 3.0 stable with the 2.4 kernel on SMP
samba servers with large (80GB IDE hardrives). We do lots of big file
(700MB) transfers. No problems yet.


==============
(Johann Spies)

I tried it out on our ftp-server (cluster) but switched over to ext3
because of instability when reiserfs and nfs when used together.


======
(Nate)

My experience with reiserfs is limited to the 2.2 series of kernels
since I do not use 2.4 on a large scale yet.

reiserfs is great for nearly all applications in my experience,
I use it in things as small as 500MB partitions on laptops to
220GB raid10 arrays. both SCSI and IDE, I use it on about 15
different servers and probably 12 different desktops/laptops.

The only downsides to it under 2.2.x that I am aware of:
- no support for software raid 1,5 under 2.2.x kernels in reiserfs
- no support for badblocks under 2.2.x if you have crappy hard disks
such as IBM 75GXP it may not be a good idea to use reiserfs.

Since reiserfs is the only stable journalling filesystem that I
am aware of for 2.2.x that is why I use it. And it is probably the
oldest linux native journalling filesystem. ext3, xfs, and jfs
are still quite new (xfs/jfs have a long history on their native
platforms).


=================
(Karsten M. Self)
Linux version 2.2.18pre21 (root@shrink) 
(gcc version 2.95.2 20000220 (Debian GNU/Linux)) #1 
Thu Jan 25 18:40:22 PST 2001

Reiserfs for 1.5 years.  No problems.  Laptop.


=================================
(P.Y. Adi Prasaja, KISS May 2002)

Saya kehilangan kepercayaaan dengan ext3. kemarin, dengan load
yang lumayan tinggi (relatif dibanding biasanya), saya menemukan
beberapa file di corrupt directory postfix. ah .. it's ok saat
itu saya pikir karena mungkin ada hubungannya dengan drweb
atau modified avcheck yang saya buat.

Kemarin malam saya coba di rumah, well dengan load yang super tinggi.
ternyata saya menemukan file di corrupt directory juga! ini
reproducable. begitu saya switch ke ext2, problem hilang.

Akhirnya, di mail server production, saya kembalikan lagi
ke ext2. well ... memang jadi tidak reliabel :-) setidaknya
saya masih bisa berharap dengan UPS.

BTW, mail server saya sudah kembali ke load lama, 25 ribu
mail per hari tanpa noticeable load :-))) Saya semakin kurang
paham kenapa dengan load yang segitu saja, orang bisa buang
2 Gigs ram hanya untuk mailserver.


=========
(Syafril)

Saya  curiga  Ext3  ada  hubungannya  dg Processor/Speed. Saat msh pakai
P5-166  dirumah, Ext3 sering crash saat loading Xserver baik utk RH, MDK
maupun Suse, tp stl upgrade ke P3 everything fine. Memang sih masih rada
keder shg /home masih pakai Ext2.

> BTW,  mail  server  saya  sudah kembali ke load lama, 25 ribu mail per
> hari  tanpa  noticeable  load  :-)))

Coba   Install/aktifkan   X  Server  Mas,  KDE 3 full akan lbh baik atau
Ximian Desktop (Gnome 2.x boleh juga), dan test :-))))


======
(PYAP)

> Nambahin fakta, ini yg saya coba-2x (coba-2x nya pemula nih) :
> - Reiserf partisi nya si MDK nggak dikenal sama RH.

Tergantung versinya. Memang tidak spt ext2/ext3 yang bisa dimajukan
dimundurkan (versinya) seenaknya.

> - Ext-3 nya RH utk partisi / membuat HDD led saya nggak mati-2x (kasus
>   di rumah) tp sistem jalan bagus.

Ambil kernel terbaru, ambil patch untuk ext3 terbaru. install sendiri ;-)

> - Ext-3  nya  RH  terkadang  membuat  link-2x  (personal  profile) di
>   Xwindows khususnya gnome crash, tdk masalah utk kde.

Jangan menghakimi sesuatu yang belum jelas benar. Hint: jangan berpedoman
pada sesuatu yang bersifat 'terkadang' atau 'kadang-kadang'.

> - ext-3  di  rh  eth0  saya  kadang-2x  died, jika komputer ditinggal
>   hidup seharian tanpa running screen saver.

Mestinya tidak ada hubungannya antara ext3 (filesystem) dengan eth0.
Setidaknya tidak secara langsung.

> - ext-3  (jounal  file) nya mdk membuat mdk running sangat pelan, tdk
>   masalah jika ext-2

Memang journaling fs (masih) lebih lambat dibandingkan dengan
non-journaling fs. Walaupun ada beberapa kondisi yang menyebabkan
journaling fs bisa lebih cepat: jumlah file dalam 1 directory banyak
sekali. Proses baca/tulis/hapus untuk file kecil-kecil dilakukan
sangat cepat (umur file pendek), process baca file yang berulang-ulang
(yang ini cache dari system/kernel bisa menghilangkan pengaruhnya).

Ngecek pengaruh cache dari VM ini bisa dengan ls directory yang isinya
cukup banyak. ls pertama akan lebih lambat dari ls ke dua ;-)




Reply to: