From: Serguei M. <mi...@ci...> - 2003-12-03 14:36:04
|
Hi, This is a follow-up to my previous message. First of all, I finally managed to find files which were located in bad sectors using badblocks and debugfs with icheck and ncheck commands. I was lucky because those files were object .o in XFree-4.3.0 build directory which I didn't use for a long time. So, after that I ran IBM/Hitachi drive fitness tool and to my surprise it reallocated all found bad sectors. Now badblocks does not show any problem with a drive. However, I'm still not sure what I should do with the drive. It is only one year old and I've never seen any hda error in a system log. I've decided to run badblocks first time just a few days ago after I tried smartmontools and noticed an error when running long offline test. Now it has 22 reallocated sectors. My major concern is load/unload technology used in these drives. In May, 2003, I wrote e-mail to Hitachi (attached) but they replied nothing. As you can see in the smartctl output, the drive already has more than 100000 load/unload cycles, or about 30% of its guaranteed 300-350 lifetime cycles. I don't know if it means that it will work just 2-3 years more... Any idea? Thank you and best regards, Serguei. ---------- Forwarded Message ---------- Subject: Re: [smartmontools-support]Patch for Hitachi DK23xx drives Date: Tue, 2 Dec 2003 06:10:21 -0800 From: Serguei Miridonov <mi...@ci...> To: Bruce Allen <ba...@gr...> Cc: sma...@li... On Monday 01 December 2003 01:32, Bruce Allen wrote: > > Well, that may be OK but after I ran a couple of tests > > with smartctl and badblocks utility, things changed > > dramatically and this troubles me... Seems I need to > > ask for another drive... > > The drive itself may be "OK" BUT you have a number of > unreadable sectors: > > 197 Current_Pending_Sector 0x0032 083 081 000 > Old_age Always - 17 > 198 Offline_Uncorrectable 0x0010 081 081 000 > Old_age Offline - 19 > > You can run the IBM/Hitachi drive fitness utility to > force these to reallocate, but data will be lost unless > the drive suceeeds in reading those sectors. The problem is that badblocks utility has found 29 bad blocks and these are stright in two rows 57242-57253 and 57398-57406, and then some more: 57478, 57550, 57622, 57694, 57766, 57767, 57838, 57839 (block size is 4096 bytes, i.e. 8 sectors). Those which are in two rows, I think, can not be reallocated, so they only can be excluded from operation by 'e2fsck -c'. Also it looks like a damaged track, though I'm not sure, of course. What do you think? I tried to find files which use these blocks on the filesystem but I don't know how to do this. dumpe2fs shows that these blocks belong to 1st group: Group 1: (Blocks 32768-65535) Backup Superblock at 32768, Group Descriptors at 32769-32769 Block bitmap at 32770 (+2), Inode bitmap at 32771 (+3) Inode table at 32772-33282 (+4) 0 free blocks, 13597 free inodes, 326 directories Free blocks: Free inodes: 19108-32704 So blocks seem to be in use. However, what is interesting is that find / -mount -type f -print0 | tar -T- --null -cvPf - > /dev/null 2> fscheck.last do not show any error! I have no idea to explain this. May I ask you two questions? How many bad sectors on a disk are still acceptable? And is there any utility which can provide more information on the ext2/3 filesystem and look for files by block number? Also, could you please look again at the smartctl output (attached) and give me some advice? Thank you. Serguei. ------------------------------------------------------- |