March 14, 200818 yr Well, I'm up on 4.3-beta1. I've barely started to use it, so cannot comment on performance or stability yet. I tried playing one of the higher bit rate ISO images and it works perfectly. With a single ISO being served, the "top" command shows the cpu is 96% idle. With a parity check running at the same time as the ISO image being served, it is between 65 and 70% idle. The reboot went smoothly. I've noticed two things so far... One is very minor, the "stop" command that used to be at /root is gone. The equivalent is: mdcmd stop The second might mean a tiny bit more work for Tom. I was looking for a file on my drives and from telnet issued a "find" command. It might not be serious, but the error detected by "find" might confuse some programs that might not be as smart and able to detect broken or inconsistent file system links. See below for the error message. Joe L. root@Tower:/# cd /mnt root@Tower:/mnt# ls disk1/ disk2/ disk4/ disk6/ disk8/ user/ disk10/ disk3/ disk5/ disk7/ disk9/ root@Tower:/mnt# find . -name U* -print find: WARNING: Hard link count is wrong for ./user/TV: this may be a bug in your filesystem driver. Automatically turning on find's -noleaf option. Earlier re sults may have failed to include directories that should have been searched. ./user/data/USB_INTERFACE_FOR_JP1.doc ./user/data/Backups(pre true-image)/C/QUICKENW/INET/localweb/iis/UNUM.GIF ... ...
March 14, 200818 yr 1) The 'stop' command was just a script as follows: #!/bin/bash # samba stop sync umount /mnt/disk1 umount /mnt/disk2 umount /mnt/disk3 umount /mnt/disk4 umount /mnt/disk5 umount /mnt/disk6 umount /mnt/disk7 umount /mnt/disk8 umount /mnt/disk9 umount /mnt/disk10 umount /mnt/disk11 umount /mnt/disk12 umount /mnt/disk13 mdcmd stop You can already tell the disk count may be wrong for some configurations & it doesn't un-mount the user shares. So rather than fix it, we just deleted it. You are welcome to put it back on your system though 2) The warning message being generated by find is fixable but may have performance implications for user shares. A workaround is to specify the -noleaf option to find. Actually an easier fix for us to simply recompile find without the LEAF_OPTIMISATION feature. I'll hold off on a decision until some more time has passed - there might be another reason to fix it "properly".
March 14, 200818 yr Here's mine. It's a lil shorter and only attempts to unmount what is mounted. So if you mouint other disks in the same location they will be unmounted too. #!/bin/bash /etc/rc.d/rc.samba stop /bin/sync for disk in /mnt/disk* do umount ${disk} done /root/mdcmd stop
March 15, 200818 yr 2) The warning message being generated by find is fixable but may have performance implications for user shares. A workaround is to specify the -noleaf option to find. Actually an easier fix for us to simply recompile find without the LEAF_OPTIMISATION feature. I'll hold off on a decision until some more time has passed - there might be another reason to fix it "properly". Ok, the proper fix will be in -beta2.
Archived
This topic is now archived and is closed to further replies.