Hi All, I backup to a Transcend JF110 USB thumb drive. It has a 25 MB/s write speed. It is very fast. It backups up 940 MB in 1 minute and 32 seconds. Except if I have been running my computer all day, in which case it takes 92 minutes. I have been rebooting before doing my backups as a workaround. I get my speed back after the reboot. Programs that I run: firefox, thunderbird, gimp, open office 3, thunar, XFce, Wine, Virtual Box 1.6.6 and 2.0.2 (Guest: XP Pro sp3: word pro, 123, approach, Acrobat 8 pro, Go To Assist Express) Many thanks, -T
Bug has no I-T number, I'm working with the requestor out of band.
Steps to reproduce: 1) start a "dump" from my primary partition (/dev/sda1) to my removable SATA (not eSata) backup drive (/dev/sdb2) 2) let "dump" run for about ten minutes (full dump takes 1hr 45min to run) 3) stop the "dump" with a <Ctrl><C> After that, a 940 MB copy from /dev/sda2 to my USB stick will drop from 1 min 32 sec to 92 minutes. (Rebooting always fixes it.)
940 MB copy from /dev/sda2 ^^^^^sdb2 Ooops
It is doing it right now. I can not remember what test you wanted me to do when the symptom appears. Would you give me a list so as next time I can take some data for you? Many thanks, -T
Hi All, Just got a posting over on CentOS forum: http://www.centos.org/modules/newbb/viewtopic.php?post_id=70121&topic_id=18798&forum=37#forumpost70121 The writer was checking out why rhel 3 and rhel 5 performed so differently on large file transfers. He found that the dirty_background_ratio and dirty_ratio were set to 50 under rhel3 and 2 and 8 respectively on rhel5. He set both rhel5's to 50 and problem solved. The is what he did: root@ddb04>cat /proc/sys/vm/dirty_background_ratio 2 root@ddb04>cat /proc/sys/vm/dirty_ratio 8 root@ddb04>echo 50 > /proc/sys/vm/dirty_background_ratio root@ddb04>echo 50 > /proc/sys/vm/dirty_ratio Mine are set to: # cat /proc/sys/vm/dirty_background_ratio 10 # cat /proc/sys/vm/dirty_ratio 40 Could this problem be solved just by changing these two values? -T
Hi All, I can not reproduce this problem since upgrading to CentOS 5.4. Did you guys fix some USB bugs in 5.4? Anyway, you can close this as "works in 5.4". (I tried, but it got mad at me.) Many, many thanks, -T