ziggy99

Is the Unraid team at all working on fixing the NFS version 3 support that was broken in 6.12.11 and onwards? I tried upgrading from 6.12.8 to 6.12.13 today, and immediately could not map my NFS shares. I had to downgrade to 6.12.10 to get it working again. Due to other issues, I am forced to use NFS version 3, so it is very problematic that this suddenly stops working. I found this information in another thread: NFS3 clients go through portmapper to access nfsd. Up to unraid 6.12.10 nfsd was registered with portmapper, so nfs3 clients worked. Starting with 6.12.11 nfsd isn't registered with portmapper (gets lost when array starts), so nfs3 clients fail to mount shares from unraid nfsd. NFS4 clients to not go through portmapper to access nfsd, this is why nfs4 clients still work with 6.12.11.

I got an Unraid server with 3 WD Red SOHO NAS 4TB hard drives as data disks and 1 WD Red Plus NAS 8 TB as parity disk. The hw specs of the server is quite good with an Intel I7-8700 and 32 GB RAM. If I test writing a large file to the server (e.g. > 5 GB), the initial speed is good. I can see that it it utilizing almost the full 1 gbps network speed, but after apprx. 2 GB of data written, the speed falls to half of that. 350-450 mpbs of network utilization. The same if I write smaller files - the initial writing is fast (> 100 MB/s), but after some time it falls to 50 MB/s. Should I not expect any better speed than this from these hard drives? Why is the initial speed good, but then it decreases? It does not seem to be related to network speed - because it does not utilize the full speed of the link.

I am considering moving 4 internally mounted disks from my Unraid server to a JBOD cabinet (Qnap TL-D800S). Can I move them externally without loosing data/parity? What is the correct procedure for doing so?

I am running an Ubuntu VM on Unraid. I am exposing my shares through "9p Mode" to the VM. Inside Ubuntu, I mount the exposed shares like this: sudo mount -t 9p -o trans=virtio "backup" /media/share/backup I have a script that uses rsync to sync the files from these mounted shares onto an internal disk drive. The issues that I am facing, is that rsync sometimes reports "cannot allocate memory": rsync: [sender] readdir("/media/share/backup/myfolder"): Cannot allocate memory (12) There are plenty of memory available on the VM, so that is no issue. The next time I run the job, it might go through as normal. But I would like to get rid of these issues. The question is: Is there some kind of memory allocation/limit to the virtio/9p as well? If yes - is there some way to increase it to avoid these errors?

Ok - thanks for clarification!

I guess "can't" is a strong word here - since I am doing it. It is very practical to browse it in Unraid Dashboard to see that the backup is running fine. Do I risk anything by doing it - except of course the need to remount it to see new files...

It does not seem to matter if I wait. I can wait one week and the files are not showing. I write to the disk from a VM in Unraid (Ubuntu), and the disk is made available to the VM using VirtIO setting in VM setup and /dev/sdb as location (Manual). In the VM itself, I map the disk as exfat file system: sudo mount -t exfat -o umask=0000,gid=1000,uid=1000 /dev/vdb1 /media/share/intbackup/ Everything works fine from within the VM. I can write to the disk and see the files OK. But when browsing from dashboard I cannot see the files that were written by the VM since I mounted it in Unraid dashboard.

Running latest unassigned disks plugin on latest Unraid. When I mount a disk and browse the disk content from the Unraid "Main" tab, I can see the files on the disk. The issue is that if I copy new files to the disk (e.g. in a VM), these new files are not shown when browsing again. I have to unmount and mount again to see the new files in the browse function. Is this as expected?

Thanks - that worked MUCH better!

I am running Win 11 in a VM on Unraid. Working great, but I have an issue with VNC., Sometimes it does not refresh all parts of screen, so even if the screen has changes, I do not always see them. E.g. in Windows explorer, I might not see all files until I drag the file explorer window forward and back a couple of times to force it to refresh. Happens both with default Unraid VNC web based client and standalone RealVNC Viewer. Any ideas on how to fix this? Using QXL driver in VM setting.

Hi! I have some strange SMB behavior when accessing a Unraid share (6.12.3) NB! This only happens on Mac. If I mount the share on Linux or Windows, everything works fine and I do not have these issues. Steps to reproduce: 1. Create file on Mac or server - e.g. echo "test" > aaa.txt 2. Do "ls aa*" in Mac terminal or browse folder in Finder 3. Observe that file is not found 4. Do "ls -la aaa.txt" and observer that file actually is found. Do "open aaa.txt" and observe that files is actually opened. So for some reason it will not list/browse file, but the file is there: steinrr@SteinsMacStudio tria % ls aa* zsh: no matches found: aa* steinrr@SteinsMacStudio tria % ls -la aaa.txt -rwx------ 1 steinrr staff 5 Sep 11 22:00 aaa.txt If I mount using NFS, the file shows with no issues and is browsable: steinrr@SteinsMacStudio tria % ls aa* aaa.txt steinrr@SteinsMacStudio tria % ls -la aaa.txt -rwxrw-rw- 1 1000 _lpoperator 5 Sep 11 22:00 aaa.txt I can also open the file - e.g. using history in Excel - even if I cannot browse the file. I checked the file - it has correct permissions and nothing special - it has the same permissions, ownership and attributes as all other files in the folder. NB! This just happens for new files, all old files existing on the shares before I probably tweaked smb-fruit.conf is showing fine. What can be wrong here? I did experiment with some fruit-settings to make SMB working better on Mac and I guess some of these might cause issues Here are my current smb-fruit.conf: root@stardustunraid:~# cat /boot/config/smb-fruit.conf vfs objects = catia fruit streams_xattr fruit:metadata = stream fruit:encoding = native fruit:posix_rename = yes Maybe there is something here that should not be set like this? How does default smb-fruit look? Best regards, Stein Rune Risa

I ended up using NFS instead of SMB. NFS is much faster and responsive on Mac than SMB. SMB was not usable from my Mac. NFS has been working fine, but there is one application that struggles a bit - so I am wondering what to do. I am also wondering if I should try TrueNAS instead. I have not been able to find any good documentation on disk shares. The Unraid documentation is very high level and lacking. How does disk share work? Will one share = one disk? So if you need more space on a share it is not possible due to it being bound to one physical disk? What about the exclusive share? What is it and how does it work?

@dlandon See my previous post and all the examples of what I tried. I tried that as well.

I tried: 192.168.1.0/24(sec=sys,rw,insecure,anongid=100,anonuid=99,all_squash, no_lock) 192.168.1.0/24(sec=sys,rw,insecure,anongid=100,anonuid=99,all_squash, local_lock) 192.168.1.0/24(sec=sys,rw,insecure,anongid=100,anonuid=99,all_squash, lock=no_lock) None of these worked: abc@AbcMacStudio nfshare % mount -o nfc,vers=4 -t nfs 192.168.1.130:/mnt/user/documents /Users/abc/nfshare/documents mount_nfs: can't mount /mnt/user/documents from 192.168.1.130 onto /Users/abc/nfshare/documents: No such file or directory mount: /Users/abc/nfshare/documents failed with 2 If I remove the lock related options - it works fine again. So it does not seem to be possible to do like this.

Thanks - where did you find this information? I have done extensive Googling without finding this. Can you please refer me to the documentation you found?

Attached is server log. NB! The shares works fine and has RW, performs well and behaves correctly. The issue is only with MS Word on Mac running with NFS version 4. Word thinks that files are locked - even if they are not. Excel, Powerpoint and other applications behaves correctly. Mounting the share using NFS version 3 resolves the issue and Word is ok again. I suspect it is because of the rewritten locking in NFS version 4. This was a big thing for NFS v4 - and it might get Word confused. When mounting NFS version 3 shares, it is possible to specify locking options (e.g. local locking) - but no such options are available on client side for version 4. It either does not except these mounting options or ignores them. So it would be good if there were some NFS service options in Unraid available, so the server parameters could be tuned. stardustunraid-diagnostics-20230823-0556.zip

I don’t think we need diagnostic here. There is nothing wrong with my Unraid server. I did some testing and if I used NFS version 3 (instead of 4), it worked fine. I know that locking implementation is totally changed in version 4, and I guess that this might be the reason why Word thinks the file is locked (even if it is not) In v3 you can disable locking using mount options on client, but in v4 this does not seem possible. You get an error if using the various lock options when mounting v4. it might seem that turning of locking must be done on server side, which is not possible (at least not easy) since Unraid has no option that can be set on the NFS service.

I have a strange issue with am Unraid server running NFS shares. Facts: * Running a fresh install of Word on a fresh Mac on a fresh Unraid server * Documents are on an NFS4 network share * The share and documents has all permissions: Read - Write - Execute * I can verify this by using "Get info" on a Word document and see that myselv, mygroup and all users has Read and Write. I can also verify directly on server. Documents are 777 (rwx for all) * Ready only is not set on document * Locked is not set on document Symptoms: * When I open a document located on the share in Word on the Mac - it displays a message that the document is read only and only gives me the option do duplicate it. I can duplicate it fine and save to the network location, but when opening the duplicated document, I get the read only message again. * No other programs are behaving this way on the same share - Excel, Powerpoint etc works fine. I can even open a Word document in a text editor and are able to save it - so if it definitely not read only * If I create a new Word document and saves it to this network share - I get the same behaviour - it gets read only * If I map the network share using samba (SMB) protocol instead of NFS, it works fine - issue is only with NFS (but I prefer NFS due to speed & reliability when there are many files) So somehow Word thinks that documents on the network share are read only - while they definitely not are read only. Any ideas?

I am setting up a new Mac and therefore also copying quite a lot of data from SMB shares on a new Unraid server. I am struggling with a specific Adobe Lightroom folder that is both large with multiple subdirectories and has a total of "thousands of files" (maybe 100k?). When trying to copy this folder to my Mac, Finder only displays "Preparing top copy" and never starts the actual copying. Using rsync on the Mac from command line is stuck on "Building file list". Is it "impossible" to copy this folder to my Mac without e.g. tar'ing it in Unraid before copying? I did some Googling and see that there are some SMB settings recommended for Mac - but what are the latest and greatest of what settings to use for 6.12.3?

Thanks - I did an rsync to verify afterwards and the copy had been done OK!

I am doing some transfer of 500 GB data to Unraid using Krusader from unassigned disks to some Unraid shares. The transfer seems to work well and the speed is good, but I see that Krusader suddenly shows the message "Stalled". The transfer is not stalled at all - it carries on well. See attached screenshot. 1. What does this "Stalled" mean? Why does it display it when it definitely is not stalled? 2. I am not connected through VNC all the time when I do the transfer. I close the browser with VNC and connects at some later point to check progress on the transfer. I guess this is fine? I do not need to be connected with VNC all the time? 3. Can I "trust the transfer"? Will Krusader tell me if some files could not be copied?

Same issue here... log is filling up with: Aug 8 21:16:07 MyServer nginx: 2023/08/09 06:16:07 [crit] 16233#16233: ngx_slab_alloc() failed: no memory Aug 8 21:16:07 MyServer nginx: 2023/08/09 06:16:07 [error] 16233#16233: shpool alloc failed Aug 8 21:16:07 MyServer nginx: 2023/08/09 06:16:07 [error] 16233#16233: nchan: Out of shared memory while allocating message of size 8894. Increase nchan_max_reserved_memory. Aug 8 21:16:07 MyServer nginx: 2023/08/09 06:16:07 [error] 16233#16233: *1133944 nchan: error publishing message (HTTP status code 500), client: unix:, server: , request: "POST /pub/devices?buffer_length=1 HTTP/1.1", host: "localhost" Aug 8 21:16:07 MyServer nginx: 2023/08/09 06:16:07 [error] 16233#16233: MEMSTORE:01: can't create shared message for channel /devices

Solved this using virtio shares.

Works perfectly now after the fix. The path is now better (no more /mnt in it), the scripts are running as expected and historic devices shows scripts nicely. Thanks for quick fix!

I am going to run a Linux VM on Unraid to be able to do cloud backup to IDrive. If I mount the Unraid shares in this Linux VM (using SMB) - how is the traffic routed? Will everything be internal traffic on the server - or will it go out and in again through my switch and then be restricted to my network speed? Is there any better way of getting access to these shares/file from my VM than to map them as SMB? Can the VM get direct access to the shares on the Unraid file system somehow?