Carl Brewer <carl%bl.echidna.id.au@localhost> writes: > The funny part to my eye us the start at 2048 not 63? Is that right? 63 is leftover from when floppies actually had 63 sectors per track. People use 2048 instead so that when sectors are really 4K rather than 512 an e.g. 4K write from the OS does not results in two RMW cycles because of misalignment. > Newfs runs, and then the drive makes some clicking noises around > midway through, the console shows : > > Feb 7 13:14:11 /netbsd: pciide1:1:0: timeout waiting for DRQ, > st=0xd0, err=0x00 > Feb 7 13:14:12 /netbsd: wd1a: device timeout writing fsbn 286776192 > of 286776192-286776255 (wd1 bn 286778240; cn 284502 tn 3 sn 35), > retrying > Feb 7 13:14:16 /netbsd: wd1: soft error (corrected) > > Is that normal? Should I ditch the drive? It seems to be going ok > for now, and it seems there's just the one error so far. I would say: not normal, yes. But before using a drive, I would 'dd if=/dev/zero of=/dev/rwd1d bs=256k' and then read the whole drive. That will break your fdisk partitions, but if it can't survive that, I do not expect you to have happy experience with the drive over the next two years. Also install smartmontools and use 'smartctl -a -t ata /dev/rwd1d' to read the status. I'm not sure how useful this is if this claims it is ok, but if it says something is wrong, almost certainly there really is something wrong.
Description: PGP signature