RE: datafiles on NFS?

Patrick, thanks for that very specific warning. This indicates to me that part of the basic test before putting any particular flavor of NFS into service is to be sure that:

1. basic autoextend functions correctly
2. autoextend failures on file system full are handled in a reasonable way.

I've always been shy of NFS from being a dinosaur when a particular NFS feature implementation on several flavors of UNIX nearly guaranteed silent corruption, but that particular problem was solved years ago, and whether or not certified or recommended for performance NFS mounts have been considered usable for quite a while now.

I suppose system/NFS combination that failed those tests could still be used if the DBA staff/sysadmin staff team could be definitively schooled to not allow autoextend for those mounts.

Again, thanks for the heads up on this issue. It seems like déjà vu from autoextend beyond the maximum writeable address without complaint from several years ago.

-----Original Message-----
From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Elliott, Patrick
Sent: Tuesday, March 11, 2008 6:19 PM
To: D'Hooge Freek; mzito_at_gridapp.com; spikey.mcmarbles_at_gmail.com; ORACLE-L Subject: RE: datafiles on NFS?

We ran into a specific problem at a previous vendor that I worked for where the datafiles were set to autoextend. When the autoextend maxed out the filesystem, the NFS mount acknowledged writes that didn't actually occur. The tablespace became corrupt. They had to restore from backups. This was about 15 months ago and it was SUSE Linux. This problem may not occur with other flavors of Linux, and they may have fixed it since then.

Pat

<snip>

--
http://www.freelists.org/webpage/oracle-l