We have CENTOS 4.X and 5.X on DELL servers. Recently we found some bad blocks happen on databases. the way we create file system is:
mke2fs -j /dev/sdXX
Does there has any good way can create file system with less bad block or avoid file system to use bad block?
Thanks.
______________________________________________________________________________________________________ 付費才容量無上限？Yahoo!奇摩電子信箱2.0免費給你，信件永遠不必刪！ http://tw.mg0.mail.yahoo.com/dc/landing
The bad blocks are hardware failure, you should start thinking about disks replacement.
mcclnx mcc wrote:
We have CENTOS 4.X and 5.X on DELL servers. Recently we found some bad blocks happen on databases. the way we create file system is:
mke2fs -j /dev/sdXXDoes there has any good way can create file system with less bad block or avoid file system to use bad block?
Thanks.
______________________________________________________________________________________________________付費才容量無上限？Yahoo!奇摩電子信箱2.0免費給你，信件永遠不必刪！ http://tw.mg0.mail.yahoo.com/dc/landing _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
mcclnx mcc wrote:
We have CENTOS 4.X and 5.X on DELL servers. Recently we found some bad blocks happen on databases. the way we create file system is:
mke2fs -j /dev/sdXXDoes there has any good way can create file system with less bad block or avoid file system to use bad block?
Replace the disks, bad blocks indicate the disk is failing, by the time you see bad blocks on the disk the disk has exhausted it's reserve of blocks that it can automatically re-map.
There is no reason really not to replace the disks.
nate
Thank you for answer. The problem is database report "corrupt block", but we even don't know which disk have that "corrupt block" since we use 15 disks Array.
--- 09/7/30 (四)，nate centos@linuxpowered.net 寫道：
寄件者: nate centos@linuxpowered.net 主旨: Re: [CentOS] how to create file system with less bad block?? 收件者: centos@centos.org 日期: 2009年7月30日,四,上午9:47 mcclnx mcc wrote:
We have CENTOS 4.X and 5.X on DELL servers.
Recently we found some bad
blocks happen on databases. the way we create
file system is:
mke2fs -j /dev/sdXX
Does there has any good way can create file system
with less bad block or
avoid file system to use bad block?
Replace the disks, bad blocks indicate the disk is failing, by the time you see bad blocks on the disk the disk has exhausted it's reserve of blocks that it can automatically re-map.
There is no reason really not to replace the disks.
nate
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
___________________________________________________ 您的生活即時通 － 溝通、娛樂、生活、工作一次搞定！ http://messenger.yahoo.com.tw/
mcclnx mcc wrote:
Thank you for answer. The problem is database report "corrupt block", but we even don't know which disk have that "corrupt block" since we use 15 disks Array.
Most RAID arrays have the ability to "scrub" the disks, to do low level scans on both the disks and the array as a whole, look to the management tools of your RAID controller(s).
nate
From: mcclnx mcc mcclnx@yahoo.com.tw
Thank you for answer. The problem is database report "corrupt block", but we even don't know which disk have that "corrupt block" since we use 15 disks Array.
You can format with '-c' to check for badblocks while formating. Or see 'man badblocks' (watch out) Nothing in /var/log/messages? If your disks are SMART, maybe try 'smartctl'
JD
mcclnx mcc wrote:
Thank you for answer. The problem is database report "corrupt block", but we even don't know which disk have that "corrupt block" since we use 15 disks Array.
most best practices for raids I've seen suggest using not more than 8 disks in a single RAID 5 or 6... if you need more disks in a single volume, build multiple raid sets then stripe them. For a database, you usually don't want to use raid5 at all, as the random write performance is poor, rather, use raid 1 or 10 (mirror, or striped mirrors).
RAID only protects against downtime in case of a drive failure, its no substitute for backups, it won't protect against application based data corruption or much of anything else.
A database "corrupt block" is likely -not- a hard drive problem, it means there's a part of one of the database files that has bad data in it. This can be caused by many things, only some of which are hardware related.. Older versions of Oracle did this occasionally, as does MySQL under various stress conditions.
-
John Doe -
John R Pierce -
mcclnx mcc -
nate -
Vaclav Mocek