If this isn't a problem that suddenly manifested itself, it might be the ide cable to the drive. I had a drive running slowly (a Maxtor, too) and just got a big improvement in it's performance by using a new cable. I got a round ATA133 cable for $4.00 and the drive's performance radically improved. I knew it was the cable, though, because the drive had been running fast in a nearly identical system with the only real difference being the cable. Also, though you probably know more about than I do, the -I flag of hdparm is pretty useful for info.
Richard --- "Karsten M. Self" <[EMAIL PROTECTED]> wrote: > I'm trying to find out why my disk performance is so slow -- 150x > worse > than a comperable system -- and what I can do to improve it. > > I've got a system with a painfully slow disk. The worse as it's > serving > as a Samba server for a small domain (ten workstations). Saturated > 802.1g WiFi is disk-starved presently. System response under any I/O > load is glacial. 'top' routinely reports load averages of from 2 to > 10+ > with 99% idle CPU. Memory utilization is generally nominal, and was > *not* an issue in the tests below, though disk performance pretty > much > precisely sucked. > > Using hdparm, I'm getting disk reads at 128 kB/sec. A comperable > drive > on a similar system rates nearly 150 times faster[1], at 18 MiB/sec. > I'd expect the slow drive to be capable of similar. 2.4.25 kernel on > x86 HW. All reports following are with the system booted single-user > and running minimal services. > > > I *am* applying hdparm options including both DMA and UDMA (either > mode > 2 or 4, not entirely sure). > > Specifically: > > # hdparm /dev/hda > /dev/hda: > > multcount = 16 (on) > IO_support = 3 (32-bit w/sync) > unmaskirq = 1 (on) > using_dma = 1 (on) > keepsettings = 0 (off) > readonly = 0 (off) > readahead = 8 (on) > geometry = 4982/255/63, sectors = 80041248, start = 0 > > > The disk is a Maxtor 54098U8 40 GiB UMDA66 drive. > > It's S.M.A.R.T capable, and I've run some additional diagnostics > using: > > - hdparm (-i -tT and no options) > - smartctl (-L -S -X -a -c -g -i -L and -v options) > - badblocks (no results when I left) > - bonnie++ (a disk testing utility -- no results when I left) > - Additional diagnostics: cpuinfo, dmesg, free, iostat, lsmod, > lspci, > and the /proc/ide/piix file. > > The results may be seen at: > > http://linuxmafia.com/~karsten/slow-drive/ > > Comparative results for a fast system similarly configured (but > different HW) is at > > http://linuxmafia.com/~karsten/slow-drive/fast-drive.tar.gz > > > I installed the ide-smart and smartsuite packages, with which I'm > pretty > unfamiliar. One result which appears interesting, though I'm not > sure > it's significant, is the output of 'smartctl -v /dev/hda', which > shows a > large value in the Raw Read Error Rate attribute's Raw Value > (smartctl > -a /dev/hda): > > > Vendor Specific SMART Attributes with Thresholds: > Revision Number: 16 > Attribute Flag Value Worst Threshold Raw > Value > ( 1)Raw Read Error Rate 0x000a 253 252 000 > 343037 > ( 3)Spin Up Time 0x0027 182 179 063 27627 > ( 4)Start Stop Count 0x0032 253 253 000 1910 > > The other drive has 0 rather than 343037. > > I see *no* CRC or other errors in /var/log/syslog or > /var/log/kern.log. > syslog is largely uninterrupted series of "MARK" entries. > > > > I do find a number of similar postings googling for 'hdparm kb/sec', > many involving Maxtor drives of various models, but no clear > resolutions. > > - Bad drive? > - Bad configuration? > - More data needed? If so, what? > - Other? > > Assistance very much appreciated. > > > Peace. > > -------------------- > Notes: > > 1. 149.36 times, if you're doing the math. > > -- > Karsten M. Self <[EMAIL PROTECTED]> > http://kmself.home.netcom.com/ > What Part of "Gestalt" don't you understand? > Rules of Spam: #0: Spam is theft. > > ATTACHMENT part 2 application/pgp-signature name=signature.asc __________________________________ Do you Yahoo!? New and Improved Yahoo! Mail - 100MB free storage! http://promotions.yahoo.com/new_mail -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
