On 12/01/2012 05:16 PM, Bob Goodwin - Zuni, Virginia, USA wrote:
This is becoming a bit confusing..... Everything that I've included in this message has been information on my "SERVER". While it appears that you've included information from a "SERVER" as well as "CLIENT". I think.....
BOX8 = SERVER BOX7 = CLIENT
Yes?
On 30/11/12 21:02, Ed Greshko wrote:
Well.... First of all /nfs4exports doesn't take up any space. They are bind mounts which is analogous to symbolic links. I'll illustrate using my nfs server.
My problem is that saving to /mnt/nfs4exports/home stores the data in the 50G root space. I did not realize that was happening until writing ground to a halt.
[egreshko@misty ~]$ df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/icy3-root 16G 11G 4.3G 71% / /dev/sda1 388M 47M 321M 13% /boot /dev/mapper/icy5-opt 9.7G 5.6G 3.6G 62% /opt /dev/mapper/icy5-syntegra 77G 41G 33G 55% /syntegra /dev/mapper/icy3-VMware 308G 231G 62G 79% /opt/VMware /dev/mapper/icy5-home 581G 323G 229G 59% /home
My system:
[root@box8 bobg]# df -h Filesystem Size Used Avail Use% Mounted on /dev/mapper/vg_box8-lv_root 50G 49G 0 100% / tmpfs 1.9G 112K 1.9G 1% /dev/shm /dev/sda1 485M 72M 388M 16% /boot /dev/mapper/vg_box8-lv_home 862G 520M 818G 1% /home
[egreshko@misty ~]$ exportfs -v /nfs4exports/syntegra 192.168.0.0/24(rw,wdelay,nohide,insecure,root_squash,no_subtree_check) /nfs4exports/myhome 192.168.0.0/24(rw,wdelay,nohide,insecure,root_squash,no_subtree_check) /nfs4exports 192.168.0.0/24(ro,wdelay,insecure,root_squash,no_subtree_check,fsid=0)
My NFS looks like this:
[root@box8 bobg]# exportfs -v /nfsexports 192.168.1.0/24(rw,wdelay,insecure,no_root_squash,no_subtree_check,fsid=0) /nfsexports/data 192.168.1.0/24(rw,wdelay,insecure,no_root_squash,no_subtree_check) /nfsexports/home 192.168.1.0/24(rw,wdelay,insecure,no_root_squash,no_subtree_check)
Except for the rw in /nfsexports that seems reasonable. /nfsexports is a, for lack of better terminology, is a "local" mount point on the NFS SERVER for NFS bind mounts. There should only be "directories" at that level. On my "SERVER".
[egreshko@misty ~]$ ll /nfs4exports/ total 16 drwxr-xr-x 158 egreshko egreshko 8192 Dec 1 09:47 myhome drwxr-xr-x 5 egreshko egreshko 4096 Jul 25 16:45 syntegra
[egreshko@misty ~]$ cd /nfs4exports/ [egreshko@misty nfs4exports]$ ll syntegra/ total 32 drwxr-xr-x 9 egreshko egreshko 4096 Nov 30 2011 iso drwxrwxr-x 6 egreshko egreshko 4096 Sep 19 08:55 linux-releases drwx------ 2 root root 16384 Apr 1 2007 lost+found
which is basically a "link" to /syntegra which looks like this....
[egreshko@misty nfs4exports]$ cd /syntegra/ [egreshko@misty syntegra]$ ll total 32 drwxr-xr-x 9 egreshko egreshko 4096 Nov 30 2011 iso drwxrwxr-x 6 egreshko egreshko 4096 Sep 19 08:55 linux-releases drwx------ 2 root root 16384 Apr 1 2007 lost+found
ALL on the SERVER
And my /etc/fstab contains the following....
/syntegra /nfs4exports/syntegra none rw,bind 0 0 /home/egreshko /nfs4exports/myhome none rw,bind 0 0
This above is on my SERVER to mount the actual file systems on /nfsexports for "exportation". But it appears you are showing your fstab on your "CLIENT".
My fstab:
[bobg@box7 ~]$ cat /etc/fstab
# # /etc/fstab # Created by anaconda on Tue Aug 7 08:37:16 2012 # # Accessible filesystems, by reference, are maintained under '/dev/disk' # See man pages fstab(5), findfs(8), mount(8) and/or blkid(8) for more info # UUID=4ce86523-2d45-4174-85e9-fd298e4516c8 / ext4 defaults 1 1 UUID=1a9d0937-2247-4cbc-b27d-4775c7453dba /boot ext4 defaults 1 2 UUID=96a666e2-73ac-4f5f-98bc-4a6a9383491f /home ext4 defaults 1 2 UUID=9cb5a7e0-c752-4563-b7f5-663980e090e3 swap swap defaults 0 0
192.168.1.8:/nfs4exports/ /mnt/BOX8 nfs4 rw 0 0 192.168.1.8:/nfs4exports/home/ /mnt/HOME1 nfs4 rw,bind 0 0 192.168.1.8:/nfs4exports/data/ /mnt/DATA1 nfs4 rw,bind 0 0 192.168.1.48:/SRVR1 /mnt/BOX48 nfs4 defaults,rw,user 0 0
First you are remote mounting /nfs4exports on a local mount point. This is perfectly OK to do. However doing that would negate the need for mounting /nfs4exports/home and /nfs4exports/data as data and home are sub-directories on the server. Also, is is probably ignored on remote mounts, the bind directive isn't utilized.
My CLIENT simply has
192.168.0.55:/syntegra /syntegra nfs4 defaults 0 0 192.168.0.55:/myhome /home/egreshko/misty nfs4 defaults 0 0
I did that since I wanted those directories mounted in different places on the CLIENT. But, I "could" have done this....
[root@meimei ~]# mount 192.168.0.55:/nfs4exports /mnt [root@meimei ~]# cd /mnt [root@meimei mnt]# ll total 16 drwxr-xr-x. 158 egreshko egreshko 8192 Dec 1 09:47 myhome drwxr-xr-x. 5 egreshko egreshko 4096 Jul 25 16:45 syntegra
So, you can see that the /syntegra partition uses 41GB and /home uses 323GB. None of that space is reflected on the / partition which has only 16GB allocated and 11GB used. If you combine that with /opt only 17GB is being used by the OS.
On the other hand, you have allocated 50GB to / and have used 49GB while /home is way under-utilized.
I suspect you have non-OS data on the / partition which can be moved to /home.
I would first start out by identifying what directory structure on / is using so much space. For example, you may have a /data directory that you've added which could be moved to /home.
Yes, I can move the files but I have to get the tree set up correctly. I will wait a while until I am fully functional before attempting to muddle through that.
I think you should review the link I sent to you.
Also, note that I've got my uids/gids consistent between systems so the actual names appear and not the numbers.