Systemd failing to mount a network share

I just installed systemd and I'm having some problems getting it to reliably mount the network shares I have listed in /etc/fstab.  In case it is relevant, I should note that I have opted to install systemd alongside the usual initscripts.  I've created two different Grub menu entries to select between the two systems during boot.
The relevant fstab entries are as follows:
//192.168.1.54/MP3z /mnt/MP3z/ cifs _netdev,password=foo,uid=foo,gid=foo,file_mode=0644,dir_mode=0755 0 0
//192.168.1.100/share /mnt/share/ cifs _netdev,password=foo,uid=foo,gid=foo,file_mode=0644,dir_mode=0755 0 0
//192.168.1.100/docs /mnt/docs/ cifs _netdev,password=foo,uid=foo,gid=foo,file_mode=0644,dir_mode=0755 0 0
This is what I've observed with respect to mount failures:
The share at 192.168.1.54 mounts every time without fail.
Usually, one of the other two shares at 192.168.1.100 will fail to mount while the second share at the same ip address will succeed.  Which of the two that fails is apparently random.
Only very occasionally will both shares at 192.168.1.100 mount successfully.
So far, I have never seen both of these shares fail to mount on the same test run.
I have tried changing the order of the fstab entries but it doesn't seem to affect things.
I've tried adding the "noauto" and "x-systemd.automount" options to my fstab entries.  This seems to make things work although with this setup I think the shares are mounted the first time they are browsed, not during starup.  With these mount options set, the system hangs on shutdown with either the messsage "Unset automount mnt-share.automount" or "Unset automout mnt-docs.automount".
I thought perhaps that this is happening because my network connection isn't being started in time to mount the shares correctly.  Though, I thought systemd handled these sorts of dependencies without any explicit configuration required.  Perhaps this has something to do with the two shares being on the same host (192.168.1.100)?  I don't have any problems with the third share at 192.168.1.54.
Can anyone shed some light?
Last edited by battlepanic (2012-08-15 19:28:05)

I had the same problem. While logged in as root, I can't use an alias or Cmd+K to connect to a server, but it works fine as any other user. I tried the link above and after reading the mount_afp man page, it describes two steps needed to mount a server. Here's what I had to put in Terminal to mount a share called Installers:
The root account does not seem to be getting past the first step when using Cmd+K or an alias to a network share, although this worked in previous versions of OSX.
This problem was not fixed in 10.7.1

Similar Messages

  • How Do I find out what app is auto-mounting a network share and stop it?

    I have a new laptop, the 13" rMBP.  I installed Steam and wanted to use the network share on my server to store the game files.  Everything went OK.  I closed Steam, disconnected the share and shutdown my laptop.  When I opened my laptop later on at work, it started to try and mount the network share.It keeps on going it.  I removed the Steam app and all files related to Steam that I could find.  I've looked at fstab, auto_home, auto_master and a large list of other files and directories and I *still* cannot find the offending file that's auto-mounting the share.  It's really frustrating.
    The share auto-mounts when I am home but not on the desktop or in finder.  I find it by looking at /Volumes/ in terminal.  If I mount the share manually under /Volumes/ it shows as Games-1 and mounts on the desktop under Games.
    Does anyone have any ideas on where I should be looking to find the offending process/file/daemon?
    Thank you!
    Trevor

    thanks for the reply but are you sure about this? What difference would it make? Why 5 apps?
    Even if it does work, which I doubt, at some point, I will have to resync ALL my apps again and surely the same thing will happen again, and I will be none the wiser. I still won't know which app it is. There must be a way of finding this out.
    Any ideas?
    What is really freaking me out is how it is connecting to my carrier (who is charging me) with Cellular data switched off.......?
    Thanks again.

  • Unable to mount a network share drive in mac os 10.7 when logged in as root

    I enabled the root account by entering sudo passwd root in the terminal window. Once the account was enabled successfully, I logged out from my admin account and logged in as root. I then attempted to mount a network share running on an Xserve and a Windows share but was unable to. I logged out as root and logged back in using my admin account and I was then able to mount both shares successfully.
    Is there a new Lion security feature that prevents the root account from mounting network shares? If so how can I get around it?
    Thanks for your help.

    I had the same problem. While logged in as root, I can't use an alias or Cmd+K to connect to a server, but it works fine as any other user. I tried the link above and after reading the mount_afp man page, it describes two steps needed to mount a server. Here's what I had to put in Terminal to mount a share called Installers:
    The root account does not seem to be getting past the first step when using Cmd+K or an alias to a network share, although this worked in previous versions of OSX.
    This problem was not fixed in 10.7.1

  • Mounting a network share via Applescript or Automator causes errors in MS Office apps

    If I connect a network share using an automated process, either using Applescript or Automator, though the share shows up on the Desktop and I can open it and view the various folders and files, MS Office apps including Powerpoint, Excel and Word are unable to open the files.   Each gives a slightly different error message.  In Excel, it looks like this:
    In Word, it's:
    In Powerpoint, it's:
    BUT, if I connect to the server share manually (using Cmnd-K in Finder), all files and folders are properly accessible.  I cannot see any differences between the shares connected by the two different means on the Desktop.  They have the same name, are the same size, etc.; in short, everything about them seems identical.  I even took screen shots of the Get Info page from one of them connected by the different methods:
    "home" connected by Automator:
    and connected manually:
    Does anyone have any ideas as to why this is happening or how to fix it?

    I had the same issue. After upgrading kernel to 3.4.5 today the cifs share mounted with original fstab settings. I believe it was caused by this bug:
    kernel changelog wrote:    The double delimiter check that allows a comma in the password parsing code is
        unconditional. We set "tmp_end" to the end of the string and we continue to
        check for double delimiter. In the case where the password doesn't contain a
        comma we end up setting tmp_end to NULL and eventually setting "options" to
        "end". This results in the premature termination of the options string and hence
        the values of UNCip and UNC are being set to NULL. This results in mount failure
        with "Connecting to DFS root not implemented yet" error.

  • Trying to resolve ntlmv errros mounting CIFS network shares via fstab

    Kernel: 3.4.2-2
    WM: Openbox
    About 6 months or so ago, which was after about a year on my current install with no issue, I began getting an ntlmv error when auto mounting samba shares at
    boot.  Everything still worked but I continued getting an error message.
    My fstab entry at that time looked like this:
    //<LAN_IP>/<share name>/ /mnt/Serverbox cifs credential=/path/to/file,file_mode=0777,dir_mode=0777 0 0
    The error I recieved looked like this:
    CIFS VFS: default security mechanism requested. The default security mechanism will be upgraded from ntlm to ntlmv2 in kernel release 3.3
    So I did what research I could on the error, found the "sec" option and discovered that adding the "sec=ntlmv2" option to my above noted fstab entry got
    rid of the error message and everything still worked perfectly; that is until this weekend.
    After upgrading both machines this weekend I noticed a new boot time error message and saw that my shares were no longer being mounted.
    relevant boot log:
    Mounting Network Filesystems [BUSY] mount error(22): Invalid argument
    Refer to the mount.cifs(8) manual page (e.g. man mount.cifs)
    relevant everything log:
    CIFS VFS: bad security option: ntlmv2
    /var/log/pacman from the the weekend's upgrade:
    [2012-06-16 13:03] Running 'pacman -Syu'
    [2012-06-16 13:03] synchronizing package lists
    [2012-06-16 13:03] starting full system upgrade
    [2012-06-16 13:10] removed dbus-python (1.0.0-1)
    [2012-06-16 13:10] upgraded linux-api-headers (3.3.2-1 -> 3.3.8-1)
    [2012-06-16 13:10] Generating locales...
    [2012-06-16 13:10] en_US.UTF-8... done
    [2012-06-16 13:10] en_US.ISO-8859-1... done
    [2012-06-16 13:10] Generation complete.
    [2012-06-16 13:10] upgraded glibc (2.15-10 -> 2.15-11)
    [2012-06-16 13:10] upgraded bison (2.5-3 -> 2.5.1-1)
    [2012-06-16 13:10] upgraded libpng (1.5.10-1 -> 1.5.11-1)
    [2012-06-16 13:10] upgraded cairo (1.12.2-1 -> 1.12.2-2)
    [2012-06-16 13:10] upgraded libwbclient (3.6.5-2 -> 3.6.5-3)
    [2012-06-16 13:10] upgraded cifs-utils (5.4-1 -> 5.5-1)
    [2012-06-16 13:10] upgraded sqlite (3.7.12.1-1 -> 3.7.13-1)
    [2012-06-16 13:10] upgraded colord (0.1.21-1 -> 0.1.21-2)
    [2012-06-16 13:10] installed pambase (20120602-1)
    [2012-06-16 13:10] upgraded pam (1.1.5-3 -> 1.1.5-4)
    [2012-06-16 13:10] upgraded libcups (1.5.3-4 -> 1.5.3-5)
    [2012-06-16 13:10] upgraded cups (1.5.3-4 -> 1.5.3-5)
    [2012-06-16 13:10] installed python-dbus-common (1.1.0-2)
    [2012-06-16 13:10] installed python2-dbus (1.1.0-2)
    [2012-06-16 13:10] upgraded dconf (0.12.1-1 -> 0.12.1-2)
    [2012-06-16 13:10] upgraded desktop-file-utils (0.19-1 -> 0.20-1)
    [2012-06-16 13:10] upgraded firefox (13.0-2 -> 13.0.1-1)
    [2012-06-16 13:10] upgraded freetype2 (2.4.9-2 -> 2.4.10-1)
    [2012-06-16 13:10] upgraded initscripts (2012.05.1-3 -> 2012.06.1-1)
    [2012-06-16 13:10] upgraded jre7-openjdk-headless (7.u4_2.2-1 -> 7.u5_2.2.1-1)
    [2012-06-16 13:10] upgraded jre7-openjdk (7.u4_2.2-1 -> 7.u5_2.2.1-1)
    [2012-06-16 13:10] upgraded jdk7-openjdk (7.u4_2.2-1 -> 7.u5_2.2.1-1)
    [2012-06-16 13:10] upgraded kdelibs (4.8.4-1 -> 4.8.4-2)
    [2012-06-16 13:10] upgraded libdrm (2.4.33-1 -> 2.4.35-1)
    [2012-06-16 13:10] upgraded libglapi (8.0.3-2 -> 8.0.3-3)
    [2012-06-16 13:10] upgraded liblrdf (0.4.0-9 -> 0.5.0-1)
    [2012-06-16 13:10] upgraded libmysqlclient (5.5.24-1 -> 5.5.25-1)
    [2012-06-16 13:10] installed khrplatform-devel (8.0.3-3)
    [2012-06-16 13:10] installed libegl (8.0.3-3)
    [2012-06-16 13:10] upgraded nvidia-utils (295.53-1 -> 295.59-1)
    [2012-06-16 13:10] upgraded libva (1.0.15-1 -> 1.1.0-1)
    [2012-06-16 13:10] upgraded mkinitcpio (0.9.1-1 -> 0.9.2-2)
    [2012-06-16 13:10] >>> Updating module dependencies. Please wait ...
    [2012-06-16 13:10] >>> Generating initial ramdisk, using mkinitcpio. Please wait...
    [2012-06-16 13:10] ==> Building image from preset: 'default'
    [2012-06-16 13:10] -> -k /boot/vmlinuz-linux -c /etc/mkinitcpio.conf -g /boot/initramfs-linux.img
    [2012-06-16 13:10] ==> Starting build: 3.4.2-2-ARCH
    [2012-06-16 13:10] -> Running build hook: [base]
    [2012-06-16 13:10] -> Running build hook: [udev]
    [2012-06-16 13:10] -> Running build hook: [autodetect]
    [2012-06-16 13:10] -> Running build hook: [pata]
    [2012-06-16 13:10] -> Running build hook: [scsi]
    [2012-06-16 13:10] -> Running build hook: [sata]
    [2012-06-16 13:10] -> Running build hook: [filesystems]
    [2012-06-16 13:10] -> Running build hook: [usbinput]
    [2012-06-16 13:10] -> Running build hook: [fsck]
    [2012-06-16 13:10] ==> Generating module dependencies
    [2012-06-16 13:10] ==> Creating xz initcpio image: /boot/initramfs-linux.img
    [2012-06-16 13:10] ==> Image generation successful
    [2012-06-16 13:10] ==> Building image from preset: 'fallback'
    [2012-06-16 13:10] -> -k /boot/vmlinuz-linux -c /etc/mkinitcpio.conf -g /boot/initramfs-linux-fallback.img -S autodetect
    [2012-06-16 13:10] ==> Starting build: 3.4.2-2-ARCH
    [2012-06-16 13:10] -> Running build hook: [base]
    [2012-06-16 13:10] -> Running build hook: [udev]
    [2012-06-16 13:10] -> Running build hook: [pata]
    [2012-06-16 13:10] -> Running build hook: [scsi]
    [2012-06-16 13:10] -> Running build hook: [sata]
    [2012-06-16 13:10] -> Running build hook: [filesystems]
    [2012-06-16 13:10] -> Running build hook: [usbinput]
    [2012-06-16 13:10] -> Running build hook: [fsck]
    [2012-06-16 13:10] ==> Generating module dependencies
    [2012-06-16 13:10] ==> Creating xz initcpio image: /boot/initramfs-linux-fallback.img
    [2012-06-16 13:11] ==> Image generation successful
    [2012-06-16 13:11] upgraded linux (3.3.8-1 -> 3.4.2-2)
    [2012-06-16 13:11] upgraded lirc-utils (1:0.9.0-16 -> 1:0.9.0-18)
    [2012-06-16 13:11] upgraded mesa (8.0.3-2 -> 8.0.3-3)
    [2012-06-16 13:11] upgraded mysql-clients (5.5.24-1 -> 5.5.25-1)
    [2012-06-16 13:11] upgraded mysql (5.5.24-1 -> 5.5.25-1)
    [2012-06-16 13:11] upgraded nvidia (295.53-1 -> 295.59-1)
    [2012-06-16 13:11] upgraded opencl-nvidia (295.53-1 -> 295.59-1)
    [2012-06-16 13:11] upgraded pango (1.30.0-1 -> 1.30.1-1)
    [2012-06-16 13:11] upgraded pcmanfm (0.9.10-1 -> 0.9.10-2)
    [2012-06-16 13:11] upgraded psmisc (22.16-1 -> 22.17-1)
    [2012-06-16 13:11] upgraded smbclient (3.6.5-2 -> 3.6.5-3)
    [2012-06-16 13:11] upgraded thunderbird (13.0-1 -> 13.0.1-1)
    [2012-06-16 13:11] upgraded udisks2 (1.94.0-1 -> 1.94.0-2)
    [2012-06-16 13:11] upgraded unrar (4.2.3-1 -> 4.2.4-1)
    [2012-06-16 13:11] upgraded virtualbox-archlinux-modules (4.1.16-1 -> 4.1.16-2)
    [2012-06-16 13:11] In order to use the new version, reload all virtualbox modules manually.
    [2012-06-16 13:11] upgraded virtualbox-modules (4.1.16-1 -> 4.1.16-2)
    [2012-06-16 13:11] upgraded xine-ui (0.99.6-5 -> 0.99.7-1)
    [2012-06-16 13:11] Running 'pacman -Syy'
    [2012-06-16 13:11] synchronizing package lists
    [2012-06-16 13:12] Running 'pacman -Syu'
    [2012-06-16 13:12] synchronizing package lists
    [2012-06-16 13:12] starting full system upgrade
    [2012-06-16 13:13] upgraded lib32-freetype2 (2.4.9-1 -> 2.4.10-1)
    [2012-06-16 13:13] upgraded lib32-gnutls (3.0.19-1 -> 3.0.20-1)
    [2012-06-16 13:13] upgraded lib32-krb5 (1.10.1-2 -> 1.10.2-1)
    [2012-06-16 13:13] upgraded lib32-libpng (1.5.10-2 -> 1.5.11-1)
    [2012-06-16 13:13] upgraded lib32-libx11 (1.4.99.902-1 -> 1.5.0-1)
    [2012-06-16 13:13] upgraded lib32-nvidia-utils (295.53-1 -> 295.59-1)
    [2012-06-16 13:13] upgraded lib32-sqlite3 (3.7.11-1 -> 3.7.13-1)
    [2012-06-16 13:13] upgraded lib32-util-linux (2.21.1-1 -> 2.21.2-1)
    [2012-06-16 13:13] upgraded lib32-xcb-util (0.3.8-1 -> 0.3.9-1)
    [2012-06-16 13:13] upgraded wine (1.5.5-1 -> 1.5.6-1)
    Currently returning to the old fstab entry once again gives the initial error code about the security mechanism being upgraded in kernal release x.x (it always seemed to change with each kernel change) though the shares seem to mount just fine. I've looked through the wiki, man pages on die.net and googled everything I can think of and I find a lot pages mentioning ntlmv errors with no solutions, many telling me that ntlmv and ntlmv2 are mount options, but nothing that gives me any indication on why I might be getting this error or how to go about looking for a solution.
    I've looked through the pacman logs on both my desktop and my file server that I'm connecting to in an effort to determine what might have changed and I found that:
    the smbclient had been upgraded on both machines so I tried downgrading back to version 3.6.5-2 but there was no change when rebooting.
    I also found cifs-utils had been upgraded on the file server.  So I downgraded that as well to the previous version (5.4-1), rebooted both machines and I'm still getting the same invalid arguement error.
    I've now gone back and upgraded to the most recent versions of the downgraded packages on each machine but I'm at a loss as to what my next steps should be.  Where do I go from here to track this down and determine if this is a bug or configuration error.  Is there a cleaner way of mounting these shares that I should be using instead of fstab?
    Thank you.

    I had the same issue. After upgrading kernel to 3.4.5 today the cifs share mounted with original fstab settings. I believe it was caused by this bug:
    kernel changelog wrote:    The double delimiter check that allows a comma in the password parsing code is
        unconditional. We set "tmp_end" to the end of the string and we continue to
        check for double delimiter. In the case where the password doesn't contain a
        comma we end up setting tmp_end to NULL and eventually setting "options" to
        "end". This results in the premature termination of the options string and hence
        the values of UNCip and UNC are being set to NULL. This results in mount failure
        with "Connecting to DFS root not implemented yet" error.

  • 10.8 fails to mount 10.5 shares

    So I just installed 10.8 last night and right off I was alerted that it couldn't mount my media and backup shares that reside on a Mac Mini running 10.5. Is anyone else noticing issues mounting shares on Macs running anything lower than 10.6? The error message being displayed is that the share is unavailable or not on. I was able to confirm all was OK from my wife's MacBook which was still running 10.7. It was able to mount them just fine.

    For not seeing how to delete a post, I thought I'd note that this is no longer an issue. Restarting seems to have corrected the problem. Not sure what the problem was originally, but all appears to be OK now.

  • How can I stop an network share from auto mounting?

    I have a recent 13" MBPr and I installed Steam.  For kicks, I told it to save the library on my 10.8 mac mini server.  This worked well.  It worked so well, I started up Parallels and did the same thing for my steam library on my boot camp partition. I closed Parallels and Steam then put my laptop to sleep.  I was away from my home network and repeatedly received messages about not being able to access my server in a standard Finder pop-up.  No big deal, I opened Steam to try and remove the library from my list.  It was no longer listed.
    Upon arriving home I opened steam and still could not find my share in the list.  I removed Steam and all it's support files.  Still on every boot I get the finder error about not being able to connect to my server.  When I boot up my parallels boot camp VM I get the same Finder error.  It pops up several (I never bothered to count) before it stops after I keep closing the pop-up.
    I've looked at /etc/auto_master and /etc/auto_home, my login items, and I cannot find the process that keeps trying to mount the network share.  When I am at my home and reboot my mac or boot parallels I do not receive the error.  But I also do not see the mounted share on my desktop.  I do see it when I perform the ls command on /Volumes.  When I try and cd into it I get a Permission Denied error.  Root/Wheel seems to own the mounted share.  When I sudo the folder I see all the files stored on that volume.
    Does anyone know how I can stop this from happening without resorting to creating a new user?  I just got this one the way I like it (before I screwed it up with steam!)
    Thanks for reading my short novel.

    did the restore through itunes but to no avail.
    Then you must have restored and installed your backup when requested.  To remove those emails, you have to restore as new, i.e. without your backup, as roaminggnome suggested.

  • How to mount network share on lion permanently

    I want to know how to mount a network share from my NAS on mountian lion.
    I have seen and tried the suggestions about using command+k or the "go connect to server" and mounting that way so the password for the share is put into the keychain, then dragging the now mounted share to the system preferences user login items.
    However this isn't satisfactory. It causes windows to pop up for each share that I have when logging in.
    What I Want is for the share to appear as a mounted hard drive in the left navigation column of the finder window. I'd also like to be able to make a shortcut to the share on my desktop (or put it in my dock bar). If it shows up under devices that is perfection okay too, just so it's there somewhere logical.
    The pop up windows simply clutter up the desktop, and if they are closed, then the share is closed too, and then there is a manual process of having to use the very tedious command+k or go connect to server method to connect again.
    I need them to be able to click and open the share. I don't need them wasting time putting in share paswords and the link over and over.. the share should just be there for them as they work through the day. And if they close the folder/window, then it should be easy to get back to it.
    This is possible on windows and I want these mounts to work the same way.
    note: my NAS uses smb

    That should work. You could try deleting the passwords in Keychain associated with it and try it again.
    Another route: Create Applescript:
    Select the network drive and get info on it. Copy the server address. Open up Applescript Editor in the Utilities folder and create a new script. Type in mount volume, add space, "paste server address within quotations", and save as application to your documents. Run it from within applescript editor to verify it works. When it asks for a passwork, save it in Keychain.
    Under system preferences/accounts/login items, click the + sign to add and select the saved application.

  • Problems with mount and unmount of network shares

    Hi,
    I use an Apple script to mount four network shares during startup of an OS 10.4.4 Powerbook. Unfortunately it seems that Tiger can't always mount them all. Sometimes it only mounts three of them, but pressing Command + K indicates that the fourth is already mounted because it's grayed out. How come I can't see it on neither the desktop nor the favorites list in a Finder window then?
    I have a Tiger installation at home with Ciscos VPN Client 4.8 installed. I use that VPN client to gain access to the company's network. When the connection is established, I mount a share (with AFP over TCP/IP) on the company's server running Win2003 and ExtremeZ-IP.
    When I close down Ciscos VPN Client, the mounted network share doesn't unmount automatically, and if I forget to toss that network drive in the trash before closing down or restarting my Mac, it hangs. Why can't it just unmount that drive automatically when the VPN connection is gone?
    Regards
    Thomas G. Madsen
    Powermac G5 (2 x 1.8 Ghz)   Mac OS X (10.4.4)  

    This may happens because broadcasts do not go through an IPSec tunnel.
    Browsing Network Neighborhood is a function of the Microsoft browsing service. Any problems are usually because the PC or master browsers do not function properly. Network Neighborhood is officially not supported. However, it works if configured correctly. Browsing Network Neighborhood works by obtaining the browse list from either a master or backup browser. This list is obtained locally on your LAN by using NetBIOS Broadcasts to locate domain browsers.
    Inorder to resove this :
    Make sure you have NetBIOS over TCP enabled on the adapter that you use to connect to the domain.
    Refer this related links:
    http://www.cisco.com/en/US/products/sw/secursw/ps2308/products_tech_note09186a0080194b4a.shtml

  • Backing up network shares

    The school I work for uses a large number of Network Home Directories for Students. Our 350 students share approximately 150 MacBook Airs and iMacs, with each student using a network account to log in.  I've load-balanced the student accounts across 4 Mac Mini servers acting as file shares (with twin 256gb SSD) and 1 aging 2009 Xserve.  In years past, the student home directories were hosted on RAID volumes, providing data redundancy in case of a drive failure. With the Minis, however, this is not the case. Each mini has 2x256gb SSD drives: one drive is reserved for OS/Apps/Server Data and the second drive will be used exclusively to handle student home directories.
    I have a fifth mac mini with a 4TB Pegasus Thunderbolt RAID 5 that I would like to use as a backup server for these Home Directories. I'd like to use some combination of rsync/automator/ical to schedule these backups to happen nightly (like, between 10:00pm and 5:00am).
    I was thinking that:
    BackUpMini would need to mount the network shares of each of the 5 file servers:
         /XServe5/Home Directories/students a-g
         /Mini1/Home Directories/students h-l
         /Mini2/Home Directories/students m-o
         /Mini3/Home Directories/students p-s
         /Mini4/Home Directories/students t-z
    BackUpMini would have the RAID mounted with a volume called "Home Directories Backup" with unique directories
         students a-g
         students h-l
         students m-o
         students p-s
         students t-z
    An rsync script would copy each source folder to its corresponding target volume on the RAID.
    Does the logic seem sound?
    If so, does anyone have any advice on what a script should look like to handle this all in one shot?
    Thanks for any feedback,
    Christopher

    So a few thoughts that may help you down the right path.
    First, if you are equally distributing your 350 students across the 5 servers as seen here:
         /XServe5/Home Directories/students a-g
         /Mini1/Home Directories/students h-l
         /Mini2/Home Directories/students m-o
         /Mini3/Home Directories/students p-s
         /Mini4/Home Directories/students t-z
    We can assume a normalization of there being 70 home folders per server.  (Obviously, the number of students in the T-Z is likely less, but we will stick with the averages).  If this is the case, you might be running a real tight ship when considering that a 256 GB disk will provide an average of only 3.65 GB per student home folder.  Not knowing the age or the work content, this could be very tight and may result in some issues, especially if you are not redirecting cache and temp files.  Consider that each home folder, without user template modification is 8.3 MB before logging in.  Once logged in, the home folder can triple without launching a single app.
    Again, you have the experience so I trust you ran the numbers.  I know however that I would be very concerned about a volume so small.
    So now to your question.  Yes, rsync can do this very effectively.  The 3.1.1 release has been very good.  Add the creation times and extended attribute patches when you compile. 
    And, assuming that you are looking to do a nightly sync, then your strategy makes sense.  Plus, you should not be concerned about bandwidth.  After all, the rsync will occur in the evening when no one is using the network.  The rsync should also push the data over ssh, not by mounting AFP/SMB shares.  Tunneled over ssh is faster, does not require the Finder to be active, and can easily be automated via cron or launchd.  Simply exchange your ssh key and your backup system will function fully automated. 
    Also, if all of the machines are integrated into the same domain, you can even push all permissions.  If you opt to do this, rsync should connect as the root user (must edit sshd.conf) so that you can traverse all the folders, avoiding permissions lockout.  In theory, if you do this right, you will be able to move the backup volume to the failed mini, remap the net home mount, and pick up right where you left off.
    And remember, the minis can be link aggregated to support more network bandwidth.  I remain a fan of the USB to 1000Base adaptors since it leaved the TB port open.  But Thunderbolt to 1000Base will work just as well.  Talk to your switch team to enable 802.3ad link aggregation on the switch and you can run the minis at 2000Base.  For the $40 cost of an adaptor, this extra speed is really a bonus.
    Oh, one other tip about rsync.  I've gotten into the habit of wrapping my rsync script into a conditional.  What I learned (the hard way) is that rsync does what you tell it and does not question.  This can be devastating...  So I've been doing my rsyncs like this and I sleep better at night:
    #!/bin/bash
    ## Base Variables
    # I define a date stamp with something like:  theDate=`date +%m_%d_%y_processName`
    # I will define a log path so I can change it easy:  baseLogPath=/Volumes/Backup/Library/Logs/
    ## Confirm that Source is Present
    ## Assuming I am triggering this from the backup server, I will first confirm that the source is visible.  If it is not, I run the risk of deleting all my content
    volumeSane=`ssh [email protected] "test -e /Volumes/Data/.rsyncTrigger && echo 1 || echo 0"`
    if [ $volumeSane = 1 ]; then
    # run the rsync only if the source volume is present
      (rsync -avX -e ssh "[email protected]:/Volumes/Data/SharedFolders/" "/Volumes/Backup/SharedFolders" --delete --stats >> ${baseLogPath}${theDate}.log) >& ${baseLogPath}${theDate}.error
    elif [ $volumeSane = 0 ]; then
    # don't run if the source is missing
    # I also like to do a osascript -e "display notification \"hey it failed\""
      echo "source volume not present -- exiting" >> ${baseLogPath}${theDate}.error
    fi
    Reid
    Apple Consultants Network
    Apple Professional Services
    Author "Mavericks Server – Foundation Services" :: Exclusively available in Apple's iBooks Store

  • Can you use a network share as location for users' home

    I am running a Mac mini with Mountain Lion and Server.app as a home server. For added storage space I have a NAS. I would like to set up the open directory network users' home share on the NAS. Is that even possible?
    I am able to mount the network share on the mini using either afp or nfs, and I can also add the NAS share as a home folder enabled share in the server.app and select that as location for a users home folder, but that only results in the user not being able to log on to his account. Is there something that I am missing or is what I am trying to accomplish simply not possible?

    It used to be officially supported to use NFS for clients to access their network home directories but with Lion (and Mountain Lion) this was no longer officially supported.
    Therefore the need as I mentioned to 're-share' the volume via AFP. As you implied this does unfortunately impose an additional network overhead as traffic has to go as follows.
    client ----> AFP ----> Server -----> NFS -----> NAS
    What you could consider to help at least a little bit is to connect the server to the NAS on a totally separate network to the network the server uses to talk to the clients on. The main network would be between the server and clients, and you would use a second Ethernet connection just link the NAS and the server. As the clients will not need to talk directly to the NAS this will not be a problem. This would at least mean that each network only gets one set of traffic either AFP or NFS but not both and means the server can use both at full speed at the same time.
    The Mac Pro still has two built-in 1Gbps Ethernet ports but you can also get either a USB3 Ethernet adaptor or a Thunderbolt to Ethernet adaptor (I have used one of these on a Mac mini server).
    If you did not already have the NAS then people starting from new would be better off either getting a Thunderbolt RAID which can be directly attached to the Mac server, or at the higher-end go the traditional route of setting up a SAN and using an FDDI connection.
    Note: You can now get Thunderbolt to FDDI interfaces.

  • SQL server can't access network share

    Hello,
    I am facing an issue where the SQL server can't access the network share. The network share is mounted by  a domain user who is also configured as Administrator on the server.
    Through SQL Server management studio as well as PoweShell script when i try to access the share in order to attach the database contained within the share, i get the error "Unable to find the file specified". I can't mention the network drive to
    the SQL management GUI file browser too if i try to attach the database through GUI.
    For all SQL services, I changed the service log on account to be the same account which mounts the network share.
    As work around - 
    I launched a command prompt under NT Authority \ SYSTEM (LocalSystem) account and mounted the drive to the same location which lead to adding one more connection but under local system account. With this change, the SQL server was able to access the network
    share!
    I couldn't understand whey SQL expects the network share to be mounted under local system account even if all the SQL services are running under domain account ? I want SQL to be able to access the network share without opting for above work around.

    Hello,
    I couldn't understand whey SQL expects the network share to be mounted under local system account even if all the SQL services are running under domain account ? I want SQL to be able to access the network share without opting for above work around.
    It doesn't have to be. The workaround is only a temporary fix and is certainly not a good idea. I would not use mounted folders with SQL Server, at all, period.
    The reason you're having difficulty is that when the folder is mounted, it's mounted for YOUR session not ALL sessions. SQL Server has already started its services and is running under its' own session and is not privy to the things you do to change the
    environment. For example, if you changed a system or user variable (for example, a SET variable) SQL Server would have no idea about this because when it started, that variable either didn't exist or was something else. This is not SQL Server's fault but how
    Windows works (this information is stored in the Process Environment Block [PEB]) for any windows process.
    As other's have said, don't use mounted drives - use a UNC path. UNC paths are 100% supported on 2012+ as well as running databases on a SMB 3.x share (UNC pathed of course).
    If you absolutely must use a mounted drive, enabled xp_cmdshell access (through sp_configure) and run the net use command through xp_cmdshell. I, once again, would highly advise not doing this and using an UNC path.
    -Sean
    The views, opinions, and posts do not reflect those of my company and are solely my own. No warranty, service, or results are expressed or implied.

  • Time Machine Can Be Made to work with a Network Share

    FYI I am just posting this here for the benefit of the community. The source of this info is: http://forums.macrumors.com/showthread.php?t=375047
    As I write this, I am doing a Time Machine backup to a ReadyNAS NV+ afp mounted share. Here is a summary of what the thread above says you need to do. It worked for me.
    Take any supported USB standalone drive and plug it in to an available USB port.
    Use Disk Utility to format the drive (or partition on the drive) and name that partition to whatever you want your network share to be called. In my case, I called it "office-tm".
    After you do this, open the Time Machine preference panel and select that new drive (in my case "office-tm") to be the Time Machine backup drive.
    After Time Machine sets up the drive, it will start counting down before the next backup. Press the little X on the right side to cancel the backup.
    Now open a terminal session and change directories to /Volumes/<drive> where in my case <drive> was office-tm (cd /Volumes/office-tm).
    When you do a ls -a -l, you will see two important files. One starts with a period (.) and contains numbers after it. The other one is called .com.apple.timemachine.supported.
    What you need to do is copy these two files to your network share. The trick is that you want your network share to have the same name as this drive you just prepared (again, in my case it was called office-tm). I first created a new share on my ReadyNAS+ called office-tm and set the quotas properly.
    Then you need to copy the files. There are several ways to do this but the way I did it was that I temporarily mounted the network share at a different mount point in terminal just to get the files over. So for example, I did:
    sudo mount -t afp afp://<ip.addr.of.share>/office-tm /Volumes/office-tm2
    Then I copied the two files over (substitute your appropriate volume names):
    cp /Volumes/office-tm/.00whateveryourfilenameis /Volumes/office-tm2/.
    cp /Volumes/office-tm/.com.apple.timemachine.supported /Volumes/office-tm2/.
    Turn off Time Machine so you can unmount.
    Now you can unmount both the USB drive and the network share and then you remount the network share as the proper name. The same one that the USB drive was named so as far as Time Machine is concerned, the volume is still there. The two files are on the network share so all is fine and Time Machine can find the volume it expects.
    Wait for Time Machine to start it's hourly backup and away you go.
    PLEASE NOTE THIS DISCLAIMER - There is probably a good reason why Apple disabled this feature so you might want to think carefully before you rely on this hack published in the MacRumors forum to back up your Mac. I am using this for testing purposes only and I am still using traditional backup techniques for my "real" backup of my office machine. If you don't get good Time Machine backups and you have relied upon them for critical data, don't blame anyone but yourself.

    Name: RAIDiator
    Version: 3.01c1-p6
    Date: Tue Dec 12 17:46:22 2006
    Size: 46566912 bytes
    Reason: Final 6.
    We should probably start a new topic in networking on this. I am seeing some strangeness with afp shares as well.
    I have most shares set up to allow access by both CIFS/SMB and AFP. If I use Sidebar and try to access the shares by AFP, it says it can't connect to the server. But the same shares via CIFS works fine. (I can click on the "Windows" server icon and connect.) But for any of these shares, I can manually connect with the normal Go > Connect to Server... method, specifying afp://<ip address> and that works fine.
    In the case of the Time Machine share that is the subject of this post, I disabled CIFS and am only using AFP because I wanted to make sure I connected to the share only via AFP for Time Machine purposes.
    BTW, I do not use USER or DOMAIN security mode in my case and my shares are set up as SHARE security mode.

  • Trouble With mouting network share

    We got this problem on 3 mac who are in 10.3.9
    When the user mount a network share, ok no problems.
    When he want to mount a second share the first one disconnect.
    the alias on the desktop dissapear, but the one on the left panel of the finder stay, and it's impossible to remount it because, the share stay on /Volumes/
    i tried to remove it => impossible sill reboot.
    i got another problem on these machines with mounting network sharing.
    When i select a share on the dropdown list, it mount the one who's before.
    i did a onyx, diskwarrior repair permission etc ...., other ideas ?
    Thank you

    No idea ?

  • Network Shares Disappear on Wake From Sleep

    I recently got a Mac mini to replace an aging Mac Pro. I copied all my data over and everything is running great, and much faster overall. But I'm having one strange bug. When I put my Mac mini to sleep, it wakes up and all of my network shares are gone. I have 4 shares I keep connected all the time, and many more I use occasionally. All of these servers are Macs. I never had an issue like this on the Mac Pro, in all it's years of service. Including the time it was running 10.8. But now, the mini dumps shares - I think every time it goes to sleep, if not, almost every time. It's a real hassle having to reconnect to each share every time you wake up. Plus it's no fun having open documents dissapear either. My network settings are pretty simple, my primary connection is gigabit ethernet. It's first in the "service order" list. I also have airport enabled but not connected. I leave it on so my computer can have location serivces. This is all exactly the same as the setup in the old Mac Pro.
    Any ideas?

    If you must put your computer to sleep, you could use an applescript to re-mount your network shares when you wake it up.
    tell application "Finder"
              mount volume "afp://servername.local/sharename" as user name aUsername with password aPassword
    end tell

Maybe you are looking for