cancel
Showing results for 
Search instead for 
Did you mean: 

Oracle + Linux64 + ReiserFS = total nightmare?

Former Member
0 Kudos

Hello All,

We have a number of boxes running on Linux64 with filsystem ReiserFS and using Oracle DB. Absolutely standart configuration. And we have a huge amount of issues with data corruption. Like Oracle can occasionaly stops with messgaes - datafile corruption, redologs get corrupted, etc. Or, they just get hunged. Today, when we did a test for XI system failover, one of controlfiles get corrupted.

Storage is fine, our heavy R/3 boxes (number of Oracle DB's more than 4TB each) running on HP900/IA64 on HP-UX use same shared storage and no issues at all.

But Linux boxes are keep us busy all the time! We tried to use -notail option without success. I read everything on OSS and Metalink - nothing I can use. SAP told that current configuration should work fine. No futher reccomendations from them.

Anybody have idea what to do next? If you're use similar configuration, please leave me a note how is it. Our coffe machines going to be broken soon because of owernight consumption!

What we running on:

Linux ######### 2.6.5-7.202.7-smp #1 SMP Tue Nov 29 14:32:53 UTC 2005 x86_64 x86_64 x86_64 GNU/Linux

LSB Version: core-2.0-noarch:core-2.0-%:graphics-2.0-noarch:graphics-2.0-x86_64

Distributor ID: SUSE LINUX

Description: SUSE LINUX Enterprise Server 9 (x86_64)

Release: 9

Codename: n/a

SUSE LINUX Enterprise Server 9 (x86_64)

VERSION = 9

PATCHLEVEL = 2

Oracle Database 10g Enterprise Edition Release 10.2.0.2.0 - 64bi

PL/SQL Release 10.2.0.2.0 - Production

CORE 10.2.0.2.0 Production

TNS for Linux: Version 10.2.0.2.0 - Production

NLSRTL Version 10.2.0.2.0 - Production

I appreciate any input. Good, bad, whatever. We running out of ideas already.

Accepted Solutions (1)

Accepted Solutions (1)

markus_doehr2
Active Contributor
0 Kudos

Check if you have applied the fix for

983230 - LOB corruption for NOCACHE LOBs in ASSM (ORA-1555/ORA-22924)

but better question: Does those boxes have all the 30+ patches as of

871096 - Oracle Database 10g: Patch sets/patches for 10.2.0?

===

What's your setting of filesystemio_options on those boxes? You open using "setall"?

===

Using -notail does only work, if the files have been

- mount using -notail

- copy from space back

Do that will ALL filesystem, where Oracle files reside on, so /oracle, data-, log- and controlfiles.

===

Did you try ext3? Just to exclude, it's really something with the hardware...

===

Just quick shots, maybe you get an idea...

--

Markus

Message was edited by:

Markus Döhr

Former Member
0 Kudos

Thank you for suggestions!

Regarding -notail : we did exactly as you mentioned, cleared FS, remounted with -notail, copied data back onto.

I didn't get what you menat here :

===

What's your setting of filesystemio_options on those boxes? You open using "setall"?

===

And I'll check first two notes tommorow, but I believe we're implemented everything that SAP reccomended.

I'm almost ready to gave up and start planning migration to ext3, but you know... having about 30 boxes to migrate including production ones with siginficant amount of data - it's not that easy. It will take days of work to do everything let alone planning and preapartion work. So, still hope we will make ReiserFS work.

markus_doehr2
Active Contributor
0 Kudos

> Thank you for suggestions!

>

> Regarding -notail : we did exactly as you mentioned,

> cleared FS, remounted with -notail, copied data back

> onto.

>

> I didn't get what you menat here :

> ===

> What's your setting of filesystemio_options on those

> boxes? You open using "setall"?

> ===

What i meant was:

sqplus / as sysdba

show parameter filesystemio_options;

That parameter impacts, how Oracle opens the data file and has a direct relation to the "-notail" mount option.

[...]

> I'm almost ready to gave up and start planning

> migration to ext3, but you know... having about 30

> boxes to migrate including production ones with

> siginficant amount of data - it's not that easy. It

> will take days of work to do everything let alone

> planning and preapartion work. So, still hope we will

> make ReiserFS work.

Yes - it should work. We use SP3 and no updated kernel. SP2 for SLES 9 has issues in various areas, did you try to update one box to SP3?

--

Markus

Answers (0)