Portal Home > Knowledgebase > Articles Database > input/output error


input/output error




Posted by Lem0nHead, 03-08-2008, 02:05 PM
hello I'm getting the feared "input/output error" on /home partition on a freebsd server this server runs RAID-1, dmesg shows: twed0: controller error - device failure (flags = 0x40) g_vfs_done():twed0s1g[READ(offset=36997054464, length=16384)]error = 5 anyone knows what's my best action here? unmount /home and fsck it? or shutdown the server and replace a drive? (but, from this error message, I couldn't guess which drive is it... also, the 3dm raid monitoring didn't find any problem, so maybe it's the controller?)

Posted by Patrick, 03-08-2008, 03:01 PM
Have you ran a smartctl against the hard drive to see if it picks up anything unusual? Short test: smartctl -t short /dev/twed0 General information: smartctl -a /dev/twed0 Edit: If those don't work, try /dev/tw0 ... Last edited by Patrick; 03-08-2008 at 03:08 PM.

Posted by TheITAdvisory, 03-08-2008, 05:46 PM
This may or may not mean anything to you, or perhaps you have done this already? top -i In the following section look for the value wa Cpu(s): 0.0% us, 0.2% sy, 0.0% ni, 99.8% id, 0.0% wa, 0.0% hi, 0.0% si If this value is above 80% or so, then this is considered high. At this point you can come to the realization that a hard disk or controller is probably on it's way out. You can also ps faux and see if any procs have the value of D which means this proc is waiting for 'something', perhaps you have a rouge proc causing you issues Read/write issues? You can also run iostat and look at your iowait values as well. I/O errors generally speaking are indeed hard drive/controller failures, however it doesn't hurt to check the basics first, just to be sure you have covered all the bases before going after heavier diagnostic means. S.M.A.R.T systems have been known to produce false negatives (at least in some of my experiences - I get e-mails every night that "My Hard drive is possibly failing", when I know for sure it's not.) I hope that I was able to help you in some way. Good luck on your diagnostics, and please keep us posted on the results. Kind Regards,

Posted by Lem0nHead, 03-09-2008, 04:59 PM
SMART seems to be fine ps/top wasn't showing anything unusual I decided to run fsck on the partition and it found and fixed some errors now everything seems to be fine, no more error messages on /var/log/messages and so on but... don't you think I should replace the HD anyway? at least AFAIK this problem just happens when badblocks are found, what may mean the HD is dying thanks



Was this answer helpful?

Add to Favourites Add to Favourites    Print this Article Print this Article

Also Read