On Tue, 2007-10-09 at 13:11 +0200, Lars Luthman wrote: Thanks everyone for the quick replies. I mounted with mount -o sync /dev/sdb1 and now a 2 minute transfer takes 8 minutes. Does anyone have a script for monitoring with iostat or vmstat while in async mode? I'm thinking a bash equivalent to the disk monitor on gkrellm which allows me to know when a transfer has completed. Cheers. -- Patrick Shirkey Boost Hardware Ltd.

On Tue, Oct 09, 2007 at 06:58:40PM +0700, Patrick Shirkey wrote: Something like this will return when five seconds have passed with no write requests being issues to the device sda: CURRENT=$(cat /proc/diskstats | grep sda\ | awk '{ print $8 }') LAST=0 while true; do LAST=CURRENT sleep 5 CURRENT=$(cat /proc/diskstats | grep sda\ | awk '{ print $8 }') if [ "$LAST" == "$CURRENT" ]; then break fi done /proc/diskstats lists a few things. For complete devices (ie, /dev/sda instead of /dev/sda1), it lists: Field 1 -- # of reads issued Field 2 -- # of reads merged, field 6 -- # of writes merged Field 3 -- # of sectors read Field 4 -- # of milliseconds spent reading Field 5 -- # of writes completed Field 7 -- # of sectors written Field 8 -- # of milliseconds spent writing Field 9 -- # of I/Os currently in progress Field 10 -- # of milliseconds spent doing I/Os Field 11 -- weighted # of milliseconds spent doing I/Os Where field n is the nth field after the device name. -- Ross Vandegrift ross(a)kallisti.us "The good Christian should beware of mathematicians, and all those who make empty prophecies. The danger already exists that the mathematicians have made a covenant with the devil to darken the spirit and to confine man in the bonds of Hell." --St. Augustine, De Genesi ad Litteram, Book II, xviii, 37

Hi, On Tue, Oct 09, 2007 at 01:11:38PM +0200, Lars Luthman wrote: It sounds like this doesn't matter much for you, but keep in mind that -o sync doesn't cause reads to be unbuffered, so don't go trying to benchmark your flash devices that way! -Forest -- Forest Bond http://www.alittletooquiet.net

Hi, On Wed, Oct 10, 2007 at 09:51:35AM +0700, Patrick Shirkey wrote: But only for writes, no? -Forest -- Forest Bond http://www.alittletooquiet.net

Not sure if this would work, but first thing I can think of is: 1) Calculate the amount of data to transfer (du command.) 2) Find out the remaining free space in the target unit (df command.) 3) Calculate the remaining free space when the copy is done. 4) Begin copy. 5) Periodically monitor the free space. HTH, Luis

On Thu, 2007-10-11 at 11:52 +0700, Patrick Shirkey wrote: if you use your own compiled code for the copy, you can open the destination files with O_SYNC. from man 2 open: O_SYNC The file is opened for synchronous I/O. Any write()s on the resulting file descriptor will block the calling process until the data has been physically written to the underlying hardware. See RESTRICTIONS below, though. this avoids having the entire mount be done with the sync option.

On Tue, October 9, 2007 12:11, Lars Luthman wrote: just a little note of warning on this last one, specially if the usb drive is of the solid-state/flash type: the mount sync option used to force every block write and, besides it used to take much longer to complete the transfer, that used to shorten the media life, it wears out you know? that was specially true when copying large files or batches of too many, too many times. I've been bitten by this one and already trashed too many sd cards when I was ripping complete movies (200MB+) for my son's gp2x :) i don't know whether this still applies to latest kernels, or to any particular filesystem, as I'm avoiding the sync mount option on removable media (usually vfat formatted) for about a year now, but I would like to be wrong ;) cheers -- rncbc aka Rui Nuno Capela rncbc(a)rncbc.org