Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: server freezes (Suse w. Oracle 9)

Re: server freezes (Suse w. Oracle 9)

From: Jan Schledermann <jan_at_schledermann.org>
Date: Sat, 13 Apr 2002 19:45:14 GMT
Message-ID: <eV%t8.296427$Fw2.7669376@nlnews00.chello.com>


We are running 6 or 7 standard SuSE 7.2 (Kernel 2.4.4) servers with Oracle 9.01 Enterprise Edition. No problem as long as you are installing 1 patch. You find it on Oracle's technet. Search on SuSE.

We did however have a problem just like what you describe caused by the most shitty &&%*$(#)Q#$) IBM harddisks. We had some 15 pieces IBM DeskStar 75GB on disk arrays running Oracle.
Some night between 00:00 - 00:30 the server cluster would suddenly blow up leaving no trace of what had happend in the log files. At first we thought that it was caused by the disk array drivers (Fasttrak 100). That wasn't the case. Then we suspected Oracle or SuSE. No problems. We found out that the servers died in the middle of the nightly backup. Later we could reproduce the problem by selecting every single record in the database. At a certain point in time we would hit a sore spot and the server would totally crash and only run again after having removed power and starting up again.
Interestingly enough formatting the disk again and checking for bad spots would not show any bad spots at all.
We then checked the disks under Windows 2000. We had excatly the same problems.
What seems to happen is tha the headds of the disks totally flips and logically kills a spot on the disk.
In our case we where having a 30GB large database which where only used sporadically by a large number of users. The chance of hitting a bad spot was actually rather small. During backup it would always happen. The only way to solve the problem was to export all data out of the database, querying around the bad disk spots, reformat the disk and create a new database and import the data again (backup and salvaged data.

By chance I came a cross a bunch of postings on the internet covering epidemic failures of 3 different IBM desks: Dekstar 30GB, 60GB, and 75GB. After some time of usage they just totally freak out. We have a 100% failure rate of these disks which have by now removed from all systems and replaced with Maxtor drives. After installing the new drives all problems went away. A US attorney has already started a class action lawsuit against IBM for the problems with these disks.

Hope this helps

Jan Schledermann

"Kjell Nannestad" <knannest_at_c2i.net> wrote in message news:S_Pt8.919$HB3.32027_at_juliett.dax.net...
> We have installed an IBM server w. Suse and Oracle 9.
> In the night the server is totally frozen. There is no response at any
open
> telnet sessions or at the console.
>
> The only solution is to take the power.
>
> There is nothing logged in the error logs, system or Oracle, and backup of
> oracle using rman was finished
> before the server locked up.
> We have experienced something similar with AIX and Oracle 8.
>
> Is there a similar feature (bug) in Suse 7.2?
> We are using raiserfs and lvm.
>
> regards
> Kjell Nannestad
>
>
Received on Sat Apr 13 2002 - 14:45:14 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US