March 1, 201412 yr Running on v5.0 here and for a long time. There is a preclear running on a new disk, but that's about it. This was a bit alarming today as I use NFS between my Ubuntu VM's that access unRAID. While performing some updates on Plex I had it refresh the library where it then proceeded to remove tons of media from its database. When I logged into the VM to check the NFS mounts, the data for several subdirectories was unreadable. Yet I could see them when logged into unRAID. After I did that the NFS client could see the directories again. This is really weird and scary to see this kind of inconsistency from NFS. Here's a good example of what happened. A show that went missing showed up like this on the client: morris@plex01:/unraid/shows/Battlestar Galactica (2003)$ ls ls: cannot access Extras: No such file or directory ls: cannot access Season 00: No such file or directory ls: cannot access Season 01: No such file or directory ls: cannot access Season 02: No such file or directory ls: cannot access Season 03: No such file or directory ls: cannot access Season 04: No such file or directory Extras Season 00 Season 01 Season 02 Season 03 Season 04 morris@plex01:/unraid/shows/Battlestar Galactica (2003)$ ls -l ls: cannot access Extras: No such file or directory ls: cannot access Season 00: No such file or directory ls: cannot access Season 01: No such file or directory ls: cannot access Season 02: No such file or directory ls: cannot access Season 03: No such file or directory ls: cannot access Season 04: No such file or directory total 0 d??? ? ? ? ? ? Extras d??? ? ? ? ? ? Season 00 d??? ? ? ? ? ? Season 01 d??? ? ? ? ? ? Season 02 d??? ? ? ? ? ? Season 03 d??? ? ? ? ? ? Season 04 I then listed that same directory on the unRAID server: root@laffy:/mnt/user/shows/Battlestar Galactica (2003)# ls -l total 7 drwxr-xr-x 1 nobody users 5800 2013-09-15 13:48 Extras/ drwxrwxrwx 1 nobody users 112 2014-02-09 03:57 Season\ 00/ drwxrwxrwx 1 nobody users 48 2014-02-09 04:48 Season\ 01/ drwxrwxrwx 1 nobody users 752 2014-02-06 06:32 Season\ 02/ drwxrwxrwx 1 nobody users 496 2014-02-07 06:46 Season\ 03/ drwxrwxrwx 1 nobody users 112 2014-02-07 08:20 Season\ 04/ I then go back to my NFS client to list the contents again: morris@plex01:/unraid/shows/Battlestar Galactica (2003)$ ls -l total 7 drwxr-xr-x 1 99 users 5800 Sep 15 13:48 Extras drwxrwxrwx 1 99 users 112 Feb 9 03:57 Season 00 drwxrwxrwx 1 99 users 48 Feb 9 04:48 Season 01 drwxrwxrwx 1 99 users 752 Feb 6 06:32 Season 02 drwxrwxrwx 1 99 users 496 Feb 7 06:46 Season 03 drwxrwxrwx 1 99 users 112 Feb 7 08:20 Season 04 I'm finding many of my subdirectories are in this state and I have no idea why. Everything is consistent now, but..., wow, what the heck is going on here?
September 29, 201411 yr Sorry for digging up this old topic, but I just ran into the same problem. Using Unraid 6.0-beta3 (I really need to upgrade!). Mounting a user share via NFS on Ubuntu I get random directories go 'missing', for example, this is on my Ubuntu system: user@gopher:/pooh/TV$ ls -l ls: cannot access Directory2: No such file or directory ls: cannot access Directory5: No such file or directory ls: cannot access Directory6: No such file or directory ls: cannot access Directory7: No such file or directory drwxrwxr-x 1 99 users 440 Jan 12 2013 Directory1 d??? ? ? ? ? ? Directory2 drwxrwxrwx 1 99 users 248 Aug 5 01:17 Directory3 drwxrwx--- 1 99 users 952 Sep 27 02:43 Directory4 d??? ? ? ? ? ? Directory5 d??? ? ? ? ? ? Directory6 d??? ? ? ? ? ? Directory7 It is mounted as : pooh:/mnt/user/TV on /pooh/TV type nfs (rw,addr=192.168.2.61) Exported on unraid as: "/mnt/user/TV" -async,no_subtree_check,fsid=100 *(sec=sys,rw,insecure,anongid=100,anonuid=99,all_squash) Syslogs on both machines show no errors. Funny thing is, if I telnet to the unraid server and then cd into each missing directory they re-appear on the nfs client side. Any ideas on how I can fix this? I've been using CIFS up to now, but due to an issue with CIFS I'm trying out NFS. Syslog, for reference: Sep 29 12:23:10 pooh emhttp: shcmd (219): /usr/local/sbin/emhttp_event svcs_restarted Sep 29 12:23:11 pooh emhttp_event: svcs_restarted Sep 29 12:23:11 pooh avahi-daemon[1307]: Service "pooh" (/etc/avahi/services/smb.service) successfully established. Sep 29 12:23:37 pooh rpc.mountd[2502]: authenticated mount request from x:684 for /mnt/user/TV (/mnt/user/TV) Sep 29 12:23:50 pooh rpc.mountd[2502]: authenticated unmount request from x:864 for /mnt/user/TV (/mnt/user/TV) Sep 29 12:24:08 pooh rpc.mountd[2502]: authenticated mount request from x:843 for /mnt/user/TV (/mnt/user/TV) Sep 29 12:35:50 pooh kernel: mdcmd (4169): spindown 5 Sep 29 12:55:50 pooh kernel: mdcmd (4170): spindown 0 Sep 29 14:14:40 pooh in.telnetd[22956]: connect from x (x) Sep 29 14:14:41 pooh login[22957]: ROOT LOGIN on '/dev/pts/1' from 'x'
December 24, 201411 yr Hi, I have the exact same problems on 5.0.6. I assume it's the old and never-really-solved "stale nfs handle"-problem. The files/directories are 'lost' to the client after some time of system-operation. This should only happen on user-shares (but using unraid on disk-shares would be pretty pointless imo). As a solution I tried setting "Tunable (fuse_remember)" to "-1" in Settings-NFS. It improves the trouble, but still my clients run into that problem, which is it especially annoying as since I upgraded my TV-server (running VDR) from a 2.x-kernel-ubuntu-machine to debian with 3.14 backport the SMB-performance is no longer enough for flawless recording/timeshift (don't know what's different in my new kernel's CIFS-support, maybe option "cache=none" is not a perfect replacement for "directio" which I used in 2.x). I'm gonna increase my unraid-RAM from 3GB to >10GB as a last resort within the next days - hope this will eventually stop nfs' forgetfulness...
Archived
This topic is now archived and is closed to further replies.