I have a 1.5TB Western Digital hard disk (WD15EADS-00R6B0) on my CentOS-5.7 server, which has become incredibly slow for some operations, eg rsync, BackupPC archive, e2fsck, although it seems to work fine for ordinary file operations, and "smartctl -a /dev/sdb" does not report any errors.
For example, running "e2fsck -p /dev/sdb5" on a 250GB partition on this disk took over 24 hours to complete, as it did on another partition.
I wonder if anyone can offer an explanation, or suggest a remedy?
Just check smart table:
smartctl -a -d ata /dev/sdX
Sometimes disk work very slowly without any visible reason. In this case just replace your disk.
On Sun, Dec 4, 2011 at 4:46 PM, Timothy Murphy gayleard@eircom.net wrote:
I have a 1.5TB Western Digital hard disk (WD15EADS-00R6B0) on my CentOS-5.7 server, which has become incredibly slow for some operations, eg rsync, BackupPC archive, e2fsck, although it seems to work fine for ordinary file operations, and "smartctl -a /dev/sdb" does not report any errors.
For example, running "e2fsck -p /dev/sdb5" on a 250GB partition on this disk took over 24 hours to complete, as it did on another partition.
I wonder if anyone can offer an explanation, or suggest a remedy?
-- Timothy Murphy e-mail: gayleard /at/ eircom.net tel: +353-86-2336090, +353-1-2842366 s-mail: School of Mathematics, Trinity College Dublin
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Timothy Murphy Sent: Sunday, 4 December, 2011 8:46 PM To: centos@centos.org Subject: [CentOS] Strangely slow disk
I have a 1.5TB Western Digital hard disk (WD15EADS-00R6B0) on my CentOS-5.7 server, which has become incredibly slow for some operations, eg rsync, BackupPC archive, e2fsck, although it seems to work fine for ordinary file operations, and "smartctl -a /dev/sdb" does not report any errors.
For example, running "e2fsck -p /dev/sdb5" on a 250GB partition on this disk took over 24 hours to complete, as it did on another partition.
I wonder if anyone can offer an explanation, or suggest a remedy?
-- Timothy Murphy e-mail: gayleard /at/ eircom.net tel: +353-86-2336090, +353-1-2842366 s-mail: School of Mathematics, Trinity College Dublin
_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On Sun, Dec 04, 2011 at 10:24:00PM +0800, Ho Chaw Ming wrote:
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
Also, is this one of the "Green" series of WD disks? Those have a 4KB sector size, not the old/traditional 512B sector size. This has performance ramifications in that you must have partitions begin on a sector boundary or performance goes to the dogs. There are a number of articles about it on various geek sites, sorry I can't point you to one offhand.
On 12/04/11 11:08 AM, fred smith wrote:
On Sun, Dec 04, 2011 at 10:24:00PM +0800, Ho Chaw Ming wrote:
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
Also, is this one of the "Green" series of WD disks? Those have a 4KB sector size, not the old/traditional 512B sector size. This has performance ramifications in that you must have partitions begin on a sector boundary or performance goes to the dogs. There are a number of articles about it on various geek sites, sorry I can't point you to one offhand.
many newer disks use this same 4k internal sectoring. the drives still *look* like they have 512 byte sectors to the outside world, but its very important for performance to get your partitions properly aligned on a 4k boundary, and ideally your file system uses 4K (or multiple of 4k) as its block size.
sadly, the default fake CHS (cylinder head sector) mapping left over from legacy PC architectures ends up with a 255 512b sector 'track', and MBR partitioning defaults to using track or cylinder boundaries, so the first partition starts at an odd location if you use legacy tools like fdisk.
On 12/04/2011 01:08 PM, fred smith wrote:
On Sun, Dec 04, 2011 at 10:24:00PM +0800, Ho Chaw Ming wrote:
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
Also, is this one of the "Green" series of WD disks? Those have a 4KB sector size, not the old/traditional 512B sector size. This has performance ramifications in that you must have partitions begin on a sector boundary or performance goes to the dogs. There are a number of articles about it on various geek sites, sorry I can't point you to one offhand.
The WDxxEADS drives have 512B sectors, _not_ 4KB sectors.
Ho Chaw Ming wrote:
I have a 1.5TB Western Digital hard disk (WD15EADS-00R6B0) on my CentOS-5.7 server, which has become incredibly slow for some operations, eg rsync, BackupPC archive, e2fsck, although it seems to work fine for ordinary file operations, and "smartctl -a /dev/sdb" does not report any errors.
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
Thanks for your response, and for the others received.
But I should have pointed out that the slow operation has only started recently, in the last week or so, after working normally since I acquired the disk a year or so ago.
Timothy Murphy wrote:
Check your bios, and make sure that you did not have IDE mode enabled and AHCI is selected .
Thanks for your response, and for the others received.
But I should have pointed out that the slow operation has only started recently, in the last week or so, after working normally since I acquired the disk a year or so ago.
To add to the mystery, I just checked with hdparm, and the speed of the disk (as measured this way) seems normal, compared with various other disks: ----------------------------------- The "slow disk" [tim@helen ~]$ sudo hdparm -t /dev/sdb /dev/sdb: Timing buffered disk reads: 168 MB in 3.00 seconds = 55.98 MB/sec ----------------------------------- Another disk on the same machine [tim@blanche ~]$ sudo hdparm -t /dev/sda /dev/sda: Timing buffered disk reads: 146 MB in 3.01 seconds = 48.57 MB/sec ----------------------------------- An external hard disk attached to the same machine [tim@helen ~]$ sudo hdparm -t /dev/sdc
/dev/sdc: Timing buffered disk reads: 54 MB in 3.03 seconds = 17.80 MB/sec -----------------------------------
I have Windows XP on another partition. I don't think I've ever used it (I always install Windows if it is not already on the machine, as I've had a couple of occasions when it has saved my life). I'll see how fast (or slow) Windows backup goes.
Those are slowish times even for a 7200rpm disk. My desktop here at home (Ubuntu) has a slow 7200 drive and hdparam reports a lot faster than that. Well, it is a Caviar "Green" drive which means that 7200 is the fastest speed but it does spin slower too.
amckay@amckay-desktop:~$ sudo !! sudo hdparm -t /dev/sda [sudo] password for amckay:
/dev/sda: Timing buffered disk reads: 244 MB in 3.01 seconds = 81.10 MB/sec amckay@amckay-desktop:~$
You have not said anything yet in this thread about defragging that drive.
I just checked your original message and your drive is the exact same as mine except yours is the 1.5 TB version and mine is 1.0.
Also - boot a live Linux CD and then from there do hdparam again and compare results, If they differ vastly at least you know it is something in your running system which is the culprit. If they are roughly the same then it is likely the drive gone bad. Though check the man page for hdparam to see if disk fragmentation will affect the results it gives.
On Sunday 04 December 2011 19:38, Timothy Murphy wrote:
The "slow disk" [tim@helen ~]$ sudo hdparm -t /dev/sdb /dev/sdb: Timing buffered disk reads: 168 MB in 3.00 seconds = 55.98 MB/sec
Another disk on the same machine [tim@blanche ~]$ sudo hdparm -t /dev/sda /dev/sda: Timing buffered disk reads: 146 MB in 3.01 seconds = 48.57 MB/sec
Here are my results with a PATA drive and a SATA drive:
[root@poontang ~]# hdparm -t /dev/hda
/dev/hda: Timing buffered disk reads: 172 MB in 3.02 seconds = 56.97 MB/sec [root@poontang ~]# hdparm -t /dev/sda
/dev/sda: Timing buffered disk reads: 300 MB in 3.01 seconds = 99.66 MB/sec
Your results are similar to what I get with my PATA drive, so I agree with Ho Chaw Ming that you should check that AHCI is selected.
-
Alan McKay -
fred smith -
Ho Chaw Ming -
Ilyas -- -
John R Pierce -
Robert Nichols -
Timothy Murphy -
Yves Bellefeuille