From: Richard E. R. <re...@at...> - 2004-12-16 00:46:16
|
This should not have happened. I regularly run against live SAN infrastructures and have never had this issue. However, there are some scenarios that may have contributed to the corruption of the DB's: * If you did not specify a specific number of sectors to size the iometer test file, it will use all available free space on the volume or file system * Oracle has the capability to grow on the fly if needed. If there was no space available and the DB was attempting to grow, that could cause corruption. * Same for transaction logs. which should be hosted on a separate volume or file system This of course would be subject to testing. You can also bounce this off the rest of the group. _____ From: Dan White [mailto:dw...@cl...] Sent: Tuesday, December 14, 2004 9:20 PM To: 're...@us...' Subject: consequences of running dynamo against a live san Richard, I'm the head of the IT department at a small marketing firm. One of my employees decided to try measuring the IO usage on one of our Oracle database systems with iometer and dynamo. When he did that we suffered a massive failure and it appears that each data file has become corrupt. Could running dynamo to simulate load on san disk groups that are hosting a live oracle database lead to corruption like this? Or is dynamo totally innocuous? Dan White |