From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3) Gecko/20030312 Description of problem: Enabling magicdev significantly slows down disk access. That is, leaving magicdev running to poll for music cds or to auto-mount data CDS, has a major impact on the performance of the harddrive. I have a Dell Inspiron 8200 laptop. P4 2.2GHz with an IBM 5400 RPM 40 gig hd. the hard drive is hda and a dvd/cdrw is hdb. Hdb is set to ide-scsi so the burner can be used. Leaving magicdev on, hdparm reports disk speeds of 14-18 meg/sec. However, the disk feels significantly slower than that. Programs like mozilla and openoffice take upwards of a minute to run. With magicdev shutoff, hdparm consistantly reports a disk speed of 26 meg/sec. Moreover, application loading is snappy. This bug/feature of magicdev is not well documented, I've had this problem for months before I accidentily stumbled upon someone else encountering the same problem and recommending to kill magicdev. Before, I assumed the kernel was not correctly dealing with my IDE controller, and i'd need to upgrade the kernel to fix disk speed (which I was giong to get around to in the summer). Version-Release number of selected component (if applicable): RedHat 9.0, magicdev-1.1.4-4 How reproducible: Always Steps to Reproduce: 1. Enable MagicDev 2. 3. Actual Results: Slow Disk Speed Expected Results: Disk speed remains constant Additional info:
I am experiencing the same problem on a Dell M50 with a Western Digital 40GB IDE drive. I'm running redhat 7.2. When executing "hdparm -tT </dev/hdX>" with magicdev running I get disk reads of 1-2MB's/sec. When I kill magicdev disk reads go up to and are consistant at 22MB's/sec. Until a better solution is found I am going to remove the magicdev package and enable automount for the cdrom and floppy devices.
The interesting question here is really the exact model of the CD-ROM involved (though that is less of a sensible question for the laptop). The basic problem here is that the CD-ROM doesn't respond properly to magicdev's probe and that screws up the bus for some period of time. (Note that if you can possibly put your CD-ROM on a different IDE channel than your hard drive, you'll be better off) *** This bug has been marked as a duplicate of 73661 ***
The root-cause might be related to https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=179573, but there is not enough data above to be sure.
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.