Spectrum

I wonder where I got that info from? I can't find it now and I'm sure it's not something I just made up.... I found the mention of Fill-up being added in the release notes, but it's not listed in the official or the un-official manual. So will the system leave free on each drive whatever min free space is set to for any allocation method or just for fill up? Also how (is?) this parameter used when allocation method is set to high water? Even Tom seems to struggle some when explaining split level It's one of those things that makes sense when you get it, but is really difficult to explain to others... Physics instructors struggle with this problem all the time

True, those will show the partitions exist, but no info about the partitions i.e. filesystem and size. Without having that info, it's just trial and error to figure out which one has the data. fdisk doesn't support GPT so even though linux knows the partitions are there, I don't see a way to the filesystem and size info without a tool that supports GPT.

Glad that worked for you. It would be nice if there were a way to view the partition information. The fact that you were able to mount the drive means the GPT and HPFS/HPFS+ drivers are built into the kernel. I may tinker with getting gdisk or parted up and running on unRAID. No promises and don't hold your breath In the mean time, if anyone has an easier solution that already works I'm all ears!

I'm not sure whether unRAID has the GPT drivers built in the kernel and gdisk/parted are certainly not in the base install so there is no way that I'm familiar with to even list the partitions of a GPT disk. I'm out of idea's here and I don't have any mac equipment to tinker with. Hopefully someone who has been down this road before can let you know.

I don't have any drives formatted with HFS or HFS+ so I can't confirm if this will work... You can use fdisk to find where the device is mapped fdisk -l|grep HPFS and if that doesn't return anything, try fdisk -l and look for something that is NOT reiserfs, ntfs, or fat Assuming that you do the above and find that the device is mapped as sdg you can try mounting it with something like mount -r -t hfsplus /dev/sdg1 /mnt/imported_files If you don't get an error message then you should be able to see your files by executing ls /mnt/imported_files/ All of that said, I'm not sure if unRAID has the kernel drivers for HFS/HFS+ compiled into the kernel so maybe someone who knows can chime in If you can confirm that this works, I'll add the instructions to the wiki.

If the media player is loading the .jpg for display and querying the .xml for data about the recording then cache dirs probably won't help. The disks won't have to spin up when the player locates the file, but they would have to spin for the player to access the file. You might be better off adding all the disks that contain media files to a spinup group so you only get hit with the disk spinup once per session. Then they will all stay spun up while your watching videos, playing music, etc. and will all spin back down after a timeout when you stop. my $.02

Sounds like you shared out the disk shares in the Export settings section which will share out all your disk mounts i.e. /mnt/disk1 /mnt/disk2 etc. If you want to use NFS for a particular user share you need to set it on a per share basis in the user shares section. See attached pic.

I just tried mounting a user share using smbmount and the perms are indeed DIFFERENT than the source. I mounted my music share in /mnt/test using mount -t cifs //localhost/Music /mnt/test and got the following results: root@Nesoi:~# ls -l /mnt/user/Music/Mu* -rwx------ 1 root root 828166 Dec 13 01:48 /mnt/user/Music/Music.md5* root@Nesoi:~# ls -l /mnt/test/Mu* -rwxr-xr-x 1 root root 828166 Dec 13 01:48 /mnt/test/Music.md5* So when that file is copied out it has the new permissions which would show the file as hidden when exposed through samba. The exact behavior you are seeing.... I dug around a little bit and can't find any info on controlling the behavior via the mount command. There may be a way, but I haven't found it yet. One solution would be to use nfs. Export the shares as nfs ( *(rw) ) then mount using mount -t nfs server:/mnt/user/Music /mnt/tmp that get's me root@Nesoi:~# ls -l /mnt/tmp/Mu* -rwx------ 1 root root 828166 Dec 13 01:48 /mnt/tmp/Music.md5* and copies with the correct permissions. Unless you can dig up a way to force samba to just export the unix permissions via the cifs driver, you may have to use nfs or set up rsync running in daemon mode.

Samba preserves the extended attributes of files so your destination is exactly the same as your source. If you are looking for a way to remove the hidden attribute from within Linux you can use the chmod command to remove the 'world execute' bit and that will remove the hidden attribute when Windows sees the file. Here is an article that shows how Samba interchanges the Unix permissions with the Dos/Windows attributes. To remove the world execute or 'un-hide' a file named foo use chmod o-x foo To recursively 'un-hide' all the jpegs in a directory you could use chmod -R o-x *.jpg There are other ways to skin this cat, but that is the best one I came up with

Glad it helped! There really is no best way to set up user shares. What works for me might be anathema to you or vice versa You can always re-org data later, but it's a royal PITA to shuffle TBs around....

Power supplies are most efficient when used in the upper range of their ability. If you have an 800w power supply that is supplying 200w of power and a 300w supply supplying 200w of power, assuming a perfect world, matched components, etc,etc. the 300w supply would be more efficient and draw less current from the grid. That said, a modern 800w might be more efficient at a 200w load than a 10 year old 300w When sizing you need to factor in worst case scenario (ie power-on spike, all drives spinning up at the same time, etc), component degradation over time, and overhead for growth. I'll defer to others who are more familiar with what's on the market in that size right now, my last build was focused on low power so I haven't looked for bigger supplies recently.

Good deal, I didn't realize mcedit was there. We'll probably get this thing just right in time for Tom to release V5 and make it all useless historical

I have been meaning to give the tutorial some sprucing up and add a little bit of info and I finally got around to it tonight. I've added the following and would appreciate any thoughts, corrections, etc. Introduction: I added a section of assumptions. I don't want it to come across condescending, but some basic knowledge is of course required. Can anyone think of assumptions made that aren't listed that should be? First Boot: I added a section to address setting a static IP when there is no DHCP server available on the network. Thoughts on this procedure? Security: Added a new section about security. Thoughts? Here is a link to the diff in case anyone had rather see things that way, but you get all the little minor edits in there as well, and I'm more worried about the content, not the formatting but if someone wants to check that as well.....

Sounds like name resolution is not working. See this wiki article for help on fixing name resolution and that should get you sorted. unMenu is a collection of awk scripts that Joe L. wrote. To stop unMenu, log into the console or via telnet and issue killall awk and unMenu will be stopped. To restart it, issue /boot/unmenu/uu Odds are once you get name resolution sorted out, things will be fine.

Ahhh I hear ya. I know there USED to be an ftp client that would allow you to log into 2 ftp servers and initiate a transfer between them without going through you as an intermediary. -> dig dig dig -> FlashFXP was it. It lists site to site transfers on the page I linked so that might be an option. Full disclosure: I haven't used it since the mid 90's and have no affiliation with the company. Freeware wise, your best bet may be to make mount points on the Ubuntu box and mount the drives from the fstab using either cifs or nfs (pick your pleasure) and setting up VNC or FreeNX. I had FreeNX set up on a Jaunty box awhile back and it worked great. Either way, she could log in via the Windows client and just open folders and drag and drop on the Ubuntu box and it would all be taken care of.

One of the easiest methods would be to log into the Ubuntu box and mount your shares. mkdir /mnt/movies mount -t cifs \\tower\movies /mnt/movies of course replace the share names and server name as appropriate for your installation then copy the files using the cp command cp /home/myusername/movies/* /mnt/movies/ then unmount the share and delete the mount point if it is no longer needed umount /mnt/movies rmdir /mnt/movies Note that that is umount with no 'n' There are other methods, but that is quick and easy.

I'm not sure about the NIC chipset used on that motherboard. The Realtek 8112L is not listed on the Hardware Compatibility List but that doesn't mean it won't work. Raj or someone else with more build experience than me will chime in and let you know. That board has passed level 1 testing so so with the caveats listed here it should be OK. Although with one PCIe x16 and one PCIe x1 you will max out at 18 drives total on the PCIe bus and will need to go to the PCI bus if you need more drives; of course if you use 2TB drives and don't use a cache drive, that's still 34TB of storage and it's more drives than that case can handle without some "modifications" so the number of drives supported may not be an issue for you. That power supply is designed for gaming so one of the 12V rails will go to the video card power adapters and the other rail will be available for drive usage capping you at 18 amps. For storage boxen single rail PSU's are usually better. The CX430 is a good choice as is the CX600. The key is look for that single rail with a high amperage rating.

Everything was working as intended, I just didn't wait long enough for the UPS to kick off. I tested once by manually initiating /etc/apcupsd/apccontrol killpower and (after creating /etc/apcupsd/powerfail ) it powered off the UPS. The UPS immediatly came right back up so I tried again by pulling the plug and everything worked just fine. System went down then UPS powered off after 90 seconds or so. When I reapplied power it came back on and the system came right back up. Thanks for the help Joe!

That seems quite excessive! What is your lease time set to on your DHCP server? The quick fix would be to use a static IP and take dhcpd out of the loop, but that's less than ideal

If you severed both wires going to the leds you did make sure there is no way for those cut wires to cross on the side closest to the center right? It would be a shame to burn out a power supply or a motherboard from a short

Nice, didn't realize that the powerdown script did that. I have those logs but the last thing shown in any of those logs is the array shutting down. There is nothing after that for the shutdown sequence. The server shut down completely (ACPI power down it did not just halt) and I waited a minute or two and the UPS never went off. I'll try again later today when the box is not in use and give it more time before I plug it back in.

I picked up a UPS (APC Back-UPS ES 550) for my unRAID box while I was out tonight. I installed apcupsd 3.14.3 from unMenu (powerdown script was already installed) and tested things out. The system sees the UPS and performs a shutdown once the timeout trigger fires, but the UPS is not shut down even though "Power Down UPS after shutdown" is set to YES in the config. This is not a huge deal and definitely not a show stopper, but has anyone seen this or have an idea on how to rectify? Related (kind of) is there anyway to to save a log of what is going on during the shutdown other than logging to a separate syslog server? That's something I could set up if I absolutely had to, but I'd rather not

Unless you have 4.6.1 as a pre-release you should if you want optimal performance. Limetech hasn't released 4.6.1 out into the wild yet....

look in the movies directory two directories up from where you were when you moved the files. If you were in /mnt/user/foo/bar when you executed the mv command the files will be in /mnt/user/movies In a POSIX environment ../ means one directory higher in the directory tree so ../../ means 2 directories higher. Make sure you take care of this pronto and do NOT reboot. If the files get copied up into the RAMFS and you shut down they will be lost

BOOO I hope they are right Like I said, I used a modified power cable and put an el cheapo DMM in series to take the measurement. Even so, I would expect the values to be within +-10% so they are still 'good enough'. I may be able to call in a favor and borrow an actual power meter from a friend in a lab at school, but it depends on what kind of research he is working on right now and if they have any spares they could let sneak out of t he lab for a day or so. If I am able to get more accurate measurements, I will update the thread. Also, Tom posted his measurements on a similar rig here and they line up pretty well with mine! It is a shame that the board is ~$200, but I looked at it as a long term investment and the IPMI is worth the boards weight in gold. I attached a monitor and keyboard to the box for the initial setup and since then haven't had to bother. I can get a console on the box, initiate power up/down from anywhere I have internet access and my VPN key