On 04/17/18 06:35, Eyal Lebedinsky wrote: From the client /etc/fstab: 192.168.1.86:/exports/home/        /mnt/test/        nfs4 defaults    0 0 Yes, this is DF from the server. [root@Box10 bobg]# mount 192.168.1.86:/exports/home/  /mnt/test 192.168.1.86:/exports/home/ on /mnt/test type nfs # mount shows this for this server: mount 192.168.1.86:/exports/home/ on /mnt/test type nfs  I guess it boils down to how do I tell NFS to store data in /home instead of "/"? I've been looking at this and can't see what to change, or maybe it can';t be fixed that way? -- Bob Goodwin - Zuni, Virginia, USA http://www.qrz.com/db/W2BOD box10 FEDORA-27/64bit LINUX XFCE Fastmail POP3

Allegedly, on or about 17 April 2018, Bob Goodwin sent: I'm not sure of which side of the equation you're approaching this from. ------------------------ The /etc/exports file on a server exports directories on its own filesystem for external clients to use. This points to where files will actually come and go from. So if you have a filing system like: /huge-storage-space-wrong /huge-storage-space-correct (Which doesn't actually matter what partition those mount points are on.) Then change the listing in the exports file, and move any existing files from one to the other, on the fileserver. ------------------------ The /etc/fstab file points to where devices get mounted onto the directory tree. If you have mounts like /dev/wrong-device pointing to /my-preferred-storage-space Simply change the device to the one you actually want. -------------------------- Likewise, with NFS clients. They'll map your exported fileserver directories onto local ones. You can map them (almost) anywhere you want to. For permanent mounts, you can set them up in /etc/fstab with the server address pointing to the local mount point. For autofs automatic mounting, the trend is to auto-mount things inside a /net directory (which you create), and when you try to access /net/name-of-your-fileserver/name-of-your-exported-filesystem, it auto- mounts it in there/ --------------------------- Which of those scenarios are you trying to work through? NB: There's a few gotchas about NFSing things about: If you export unusual directories, or mount things in unusual places, you may have to contend with SELinux shenanigans (including bizarre unable-to-do-something kind of errors without an explanation that points to SELinux). It's usual that the numerical user IDs (and group IDs) of your users have to be the same on server and client. Yes, NFSv4 supposedly can map "tim" between user 1023 on one PC and user 1075 on another, but that doesn't always work without a configuration fight. -- [tim@localhost ~]$ uname -rsvp Linux 4.15.10-200.fc26.x86_64 #1 SMP Thu Mar 15 17:14:41 UTC 2018 x86_64 Boilerplate: All mail to my mailbox is automatically deleted. There is no point trying to privately email me, I only get to see the messages posted to the mailing list. Evolution keeps on telling me that it's refreshing, but I still want to go and get a drink.

On 04/18/2018 08:43 AM, Bob Goodwin wrote: Uhm, that looks like "box86" either isn't in DNS or /etc/hosts so it can't be resolved. If this is on the server, try "showmount -e" or "showmount -e localhost". We don't know what your server's /etc/exports line that exports the directory looks like, so there's no way to tell. The line should be something like: /home/exports/home 192.168.1.0/24(rw,no_root_squash) meaning that /home/exports/home is exported with read/write privileges to all hosts on the 192.168.1.* network, and that if a client mounts it as the root user, the root user ID is NOT squashed to the "anonymous" user on the server. You could run showmount -e 192.168.1.86 on the client to see what the server is exporting. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - To err is human. To forgive, a large sum of money is needed. - ----------------------------------------------------------------------

On 04/18/2018 11:23 AM, Bob Goodwin wrote: By default, Fedora uses the LVM (logical volume manager) system to partition the disks. It actually creates regular partitions as a raw volumes (PVs or "physical volumes"). It then typically creates a VG (volume group) that has that PVs in it. From there, it carves out LVs (logical volumes). On my laptop, for example, I have these (among others) from the "df -h" command: /dev/mapper/vg_golem4-lv_root 426G 214G 208G 51% / /dev/sda1 477M 206M 242M 46% /boot /dev/mapper/vg_golem4-lv_home 252G 49G 191G 21% /home You can see I have a /dev/sda1 partition that is used as my boot volume (/boot). Note also that my / and /home filesystems are on LVM. To see how that's set up: [root@golem4 ~]# vgdisplay -v --- Volume group --- VG Name vg_golem4 System ID Format lvm2 Metadata Areas 1 Metadata Sequence No 6 VG Access read/write VG Status resizable MAX LV 0 Cur LV 3 Open LV 3 Max PV 0 Cur PV 1 Act PV 1 VG Size 698.12 GiB PE Size 32.00 MiB Total PE 22340 Alloc PE / Size 22340 / 698.12 GiB Free PE / Size 0 / 0 VG UUID V3EZ9p-3wxH-1LJ8-ho77-Rmbf-A4d0-0oLCY7 --- Logical volume --- LV Path /dev/vg_golem4/lv_swap LV Name lv_swap VG Name vg_golem4 LV UUID lK3HOt-a76V-faDd-3Mfl-mBxZ-DTuG-gb6OqM LV Write Access read/write LV Creation host, time , LV Status available # open 2 LV Size <9.72 GiB Current LE 311 Segments 1 Allocation inherit Read ahead sectors auto - currently set to 256 Block device 253:1 --- Logical volume --- LV Path /dev/vg_golem4/lv_home LV Name lv_home VG Name vg_golem4 LV UUID GfSiWV-IpHe-HtgD-PfjA-GBuG-G3tD-eKK69c LV Write Access read/write LV Creation host, time , LV Status available # open 1 LV Size 256.00 GiB Current LE 8192 Segments 1 Allocation inherit Read ahead sectors auto - currently set to 256 Block device 253:2 --- Logical volume --- LV Path /dev/vg_golem4/lv_root LV Name lv_root VG Name vg_golem4 LV UUID rxEwZY-8BDl-zm2b-XeBh-Drqr-3Ci3-bg5JX4 LV Write Access read/write LV Creation host, time , LV Status available # open 1 LV Size <432.41 GiB Current LE 13837 Segments 2 Allocation inherit Read ahead sectors auto - currently set to 256 Block device 253:0 --- Physical volumes --- PV Name /dev/sda2 PV UUID Qasi0A-L5V4-J4EU-0D5L-fITP-BpDp-6xAash PV Status allocatable Total PE / Free PE 22340 / 0 When you look at that, you can see there's a VG (volume group) called "vg_golem4". That volume group is split up into three logical volumes, "lv_root", "lv_swap" and "lv_home" and you can see the /dev names they're known by. You can also see at the bottom that the VG has a single PV, /dev/sda2. Now as far as your /etc/exports file goes, you can have as many lines in it as you want. To wit (from an NFS server in our datacenter whose DNS name is "nfssrv598-r1"): [root@nfssrv598-r1 ~]# cat /etc/exports # /etc/exports # # Storage from HP 9320 Array (volume group "VG_HP9320")... # /adcorp 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /adlab 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /back1 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0100 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0103 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0104 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0105 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0106 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fs0107 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /fssprod 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /scratch 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) /storage 192.168.60.*(rw,no_root_squash) 192.168.69.*(rw,no_root_squash) "/adcorp" and the like are actually LVs on that server: root@nfssrv598-r1 ~]# mount | grep adcorp /dev/mapper/VG_HP9320-UTIL on /adcorp type xfs (rw,relatime,attr2,inode64,noquota) Very similar to your stuff. And from a client querying that server: [root@ing1-r1 ~]# showmount -e nfssrv598-r1 Export list for nfssrv598-r1: /storage 192.168.69.*,192.168.60.* /scratch 192.168.69.*,192.168.60.* /fssprod 192.168.69.*,192.168.60.* /fs0107 192.168.69.*,192.168.60.* /fs0106 192.168.69.*,192.168.60.* /fs0105 192.168.69.*,192.168.60.* /fs0104 192.168.69.*,192.168.60.* /fs0103 192.168.69.*,192.168.60.* /fs0100 192.168.69.*,192.168.60.* /back1 192.168.69.*,192.168.60.* /adlab 192.168.69.*,192.168.60.* /adcorp 192.168.69.*,192.168.60.* Hope that helps. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - Polygon: A dead parrot (With apologies to John Cleese) - ----------------------------------------------------------------------

On 04/18/2018 12:42 PM, Bob Goodwin wrote: No, /etc/exports tells the NFS server which directories to export via NFS. If you want to export /dev/fedora00/home, first find out where it's mounted on your NFS server. You can do that by running "df -h" as root on your NFS server. Example (again from my laptop): [root@golem4 ~]# df -h Filesystem Size Used Avail Use% Mounted on ... /dev/mapper/vg_golem4-lv_home 252G 49G 191G 21% /home So /dev/mapper/vg_golem4-lv_home is mounted on /home. So, if I were going to make my laptop an NFS server, I would add a line such as: /home 192.168.1.0/24(rw,no_root_squash) in my laptop's /etc/exports file. IMPORTANT: You export where the filesystem is _mounted_--NOT the raw device (in this case, you export "/home" and NOT "/dev/mapper/vg_golem4-lv_home"). Once that's done, I'd have to tell the NFS server daemon on my laptop that I had made changes to the /etc/exports file by running the command: [root@golem4 ~]# exportfs -ar The NFS server daemon on my laptop would then refresh its list of what it's supposed to export. To verify it, you could run a showmount -e localhost on the server to see what it thinks it's exporting, or showmount -e <NFS-servername-or-IP-address> on one of the NFS clients to show what the server is exporting. Note that the showmount command only shows the export name and not what the raw device is on the NFS server. In fact, the NFS server daemon on the server (which is what's responding to the showmount queries) doesn't even know what device holds the directory being exported. It only understands directories. Then you'd mount the export on the client. Using the above stuff, an NFS client that wanted to use my exported "/home" directory and mount it at "/nfs/home" would do something like: [root@nfs-client]# mkdir -p /nfs/home [root@nfs-client]# mount -t nfs <NFS-servername>:/home /nfs/home where "<NFS-servername>" is either the hostname of the NFS server or its IP address. The first command creates the mountpoint on the client if it doesn't already exist, the second mounts the filesystem via NFS. If you wanted to put that in your /etc/fstab ON THE CLIENT so it'd mount at boot, you'd add a line such as: <NFS-servername>:/home /nfs/home nfs defaults 0 0 Again, where "<NFS-servername>" is either the hostname of the NFS server or its IP address. Note that with that line in your /etc/fstab on the client, you could mount the filesystem via a simple [root@nfs-client]# mount /nfs/home and the mount command would look for a line in /etc/fstab that matched the filesystem specified, see that it's an NFS mount (via the "nfs" filesystem type part of the entry or the fact that the device string contained a ":/" sequence of characters which are unique to NFS volumes) and invoke the proper command. Remember, entries in /etc/exports ON THE SERVER tell the SERVER what _directories_ to export to the clients--NOT the raw device containing the filesystem. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com - - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - - - - Is that a buffer overflow or are you just happy to see me? - ----------------------------------------------------------------------

On 19/4/18 3:45 am, Rick Stevens wrote: From a subsequent mail in this thread it seems to me that Bob has the following in /etc/exports (using Rick's syntax above):                 /exports/home 192.168.1.0/24(rw,no_root_squash) which seems to me to be providing network access to the exports/home directory on whatever device in the server is the "/" mount point. If this is the case it seems to me that /etc/exports should have a specification like Rick's, assuming directory /home/exports/home exists on the server, and then the client should have the follow in /etc/fstab:                 192.168.1.86:/home/exports/home/ /mnt/test/        nfs4    defaults    0 0 or am I missing something? regards, Steve > > meaning that /home/exports/home is exported with read/write privileges > to all hosts on the 192.168.1.* network, and that if a client mounts it > as the root user, the root user ID is NOT squashed to the "anonymous" > user on the server. You could run > > showmount -e 192.168.1.86 > > on the client to see what the server is exporting. > ---------------------------------------------------------------------- > - Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com - > - AIM/Skype: therps2 ICQ: 22643734 Yahoo: origrps2 - > - - > - To err is human. To forgive, a large sum of money is needed. - > ---------------------------------------------------------------------- > _______________________________________________ > users mailing list -- users(a)lists.fedoraproject.org > To unsubscribe send an email to users-leave(a)lists.fedoraproject.org

On 04/18/18 20:03, Gordon Messmer wrote: I assume that it is NFSv4 but I don't see where that is indicated, it is recently installed ... Rick Stevens suggested: "I would add a line such as: /home 192.168.1.0/24(rw,no_root_squash)" and I was able to get it connected and working with: /home 192.168.1.0/24(rw,no_root_squash) # cat /etc/exports /home    192.168.1.0/24(rw,no_root_squash) #/exports/home 192.168.1.0/24(rw,sync,insecure,no_root_squash,no_subtree_check,fsid=0) # /exports   192.168.1.0/255.255.255.0(ro,sync) # /exports/home/public/192.168.1.0/255.255.255.0(rw,sync) # /exports 192.168.54.0/255.255.255.0(ro,sync) And the client connects with: # mount 192.168.1.86:/home/exports /mnt/testb That allows me to save data by shuffling between the files I want and putting them into /mnt/testb using the Thunar file manager, quite convenient for what I need to do. So I am making progress ... Thanks to all who responded, Bob -- Bob Goodwin - Zuni, Virginia, USA http://www.qrz.com/db/W2BOD box10 FEDORA-27/64bit LINUX XFCE Fastmail POP3

On 04/17/2018 02:42 AM, Bob Goodwin wrote: If you are using NFSv4, this is trivial.  Make a directory, /home/exports, and use that as your NFS root.  Create /home/exports/var/ftp/pub and /home/export/home/public and move the required content to those locationos.  In /etc/exports: /home/export 192.168.1.0/255.255.255.0(ro) /home/export/var/ftp/pub  192.168.1.0/255.255.255.0(ro) /home/export/home/public  192.168.1.0/255.255.255.0(rw) /home/export 192.168.54.0/255.255.255.0(ro,sync) /home/export/var/ftp/pub 192.168.54.0/255.255.255.0(ro,sync) If you want to export those directories via NFSv3, you'd create the same directory structure, and then bind mount those directories into place.  Create the directories, move the content, and then add to /etc/fstab: /home/export/var/ftp/pub /var/ftp/pub none bind 0 0 /home/export/home/public /home/public none bind 0 0 After setting up fstab, you can "mount /var/ftp/pub" and "mount /home/public".  Those will be mounted automatically each time you reboot. In that case, /etc/exports should look like: /var/ftp/pub  192.168.1.0/255.255.255.0(ro,mountpoint) /home/public  192.168.1.0/255.255.255.0(rw,mountpoint) /var/ftp/pub 192.168.54.0/255.255.255.0(ro,sync,mountpoint)