File system becoming read-only

[Nutter]

This has happened to me twice now, yesterday and 3 or 4 months ago. I've got a file server running Mandrake 10 and Samba - among other things. From my XP machine I wasn't able to write any files so I SSH'd over to the server and tried to create a file on /home/. It told me that it couldn't create because it was a read-only file system. When I switched my KVM switch over to the server I saw all types of bad block errors. Like last time I have rebooted and the boot process ran a check, found bad blocks, and is now 'invoking duplicate block passes'.

Is this a sign of a failing hard drive, or does this just happen occassionally when a server has been on for a while? Or something else?

[ChadSEO]

Nutter,

All hard drives have bad blocks. If there are only a few, then usually data is simply saved on certain blocks that are hidden just for that purpose. If you are seeing a lot of bad block errors, that is usually a sign the hard drive is on the verge of doing something bad. I would backup everything now, and start pricing out a new drive.

Chad

[Nutter]

There were 2. I suppose the real question is - is there any type of daily, weekly, monthly, whatever maintenence I should be doing instead of just fixing it when I can't write to the drive anymore?

[jtara]

If you are growing bad blocks, you should replace the drive. Period. Drives normally ship with some bad blocks, due to normal defects in the manufacturing process. But when blocks go bad after manufacture, it's an indication that the drive is failing. It is NOT normal to grow bad blocks.

Further, if your OS is seeing bad blocks, this is particularly bad.

Most drives today manage bad blocks internally and transparently. There is some special software (depending on drive type) that will allow you to view both the original and grown defect list. (grown = after manufacture). These are NOT the bad blocks that your OS is reporting to you, as your OS is completely unaware of the internal bad block list.

If your OS is reporting bad blocks, it means that either you have a (very) old drive that doesn't internally remap bad blocks, or else the drive has exhaused it's pool of spare blocks.

NOT good!

[Nutter]

Oh boy, another project :-/

[freeflight2]

check /var/log/messages* - usually the drive goes bad (or if you are lucky just the cable is not connected correctly etc.), writes fail and ext3 switches to "read-only" mode.
Test your drive with smartctl -a /dev/sdx