Posts by jbrizz

I have Transmission configured to save files to a UinionFS, suddenly it's unable to write to the downloads share anymore. I've reapplied ACLs but that made no difference. There is free space on all disks in the pool.

If I create a share on a different mount point (e.g. mdraid mirror) it works again, so it's just UnionFS. I can copy files to the downloads share over SMB as my own user account too, the UnionFS is writable.

Is someone please able to assist in getting this working again?

Thanks!

Edit: I've got it working by doing a chmod 777 on the downloads folder. Don't know why reapplying the ACLs didn't fix the problem so still curious if anyone has ideas about that. I wonder if somehow Transmission is running under a different user context, but not sure how to tell.

Looks like RPC authentication is required now, transmission is no longer working with authentication turned off. If you have this problem enable authentication and it should work again.

I guess that when I updated I got the patch to fix this vulnerability which is to force people to use authentication.

I updated my OMV 3 with latest patches in update management, one patch was for Tranmission. Since then Transmission isn't working.

The issue appears to be with the RPC part of it. The web server still serves up the Tranmission page but it gives me an error saying cannot connect to server. Remote GUIs do not work at all.

Not sure where to look. Anyone struck this? Thanks.

I have OpenVPN configured on my OMV NAS for access to my home network and anything connected to my home network. It's very easy to setup with an OMV plugin and has an Android app, Windows client and native support in most Linux distros. I connect from all three and it works perfectly.

I don't access my media remotely (although it would work over OpenVPN if I'm connected via a sufficiently fast connection) but I have the Plex port open so my parents can stream media from my house. Plex just needs the port open, remote access enabled on the server side and then you can grant access to an email address associated with a plex.tv account. You don't need sub for their premium service if you are streaming to Chromecast, smart TV or PC, I think only for phone. I have two accounts setup and both allow for unlimited streaming from TV/PC devices.

Little bit late to the party but my system is i3 6100/8GB and it can concurrently transcode two Plex streams (possibly more but never tried), torrent at few hundred mbps, run SMB for media and IP cam video, web server, run Crashplan backups etc with CPU to spare. RAM generally sits around 3 - 3.5GB used.

Hi all

I've noticed that "fast" preallocation in Transmission, which should allocate space for new downloads, does not affect total free space on the volume. If looking at the file it is the full size, but free space only goes down as download progresses. This causes an issue with UnionFS where if a few large downloads are started at once the disk will fill up without UnionFS putting any files on empty ones.

Obviously this can be fixed by setting preallocation to "full" but this is very slow to get going when downloads often have 25+ GB files.

I'm guessing this is an EXT4 or UnionFS issue, but not sure if it's by design or not. I don't see the point in fast allocation over none if it doesn't reserve the space though.

Is there a setting that can correct this or should I stick with full preallocation?

Thanks.

Thank you so much, that's exactly what I needed!

I have a share setup for downloads and the permissions are set such that the transmission user is the owner, the transmission group is the group, each of these has rwx, and other users has only rx. If I set the recursive switch when I apply the permissions everything is correct, but when a file is downloaded (created) the only account with permission to write is the debian-transmission user, not the group.

My user account is in the debian-transmission group so I think all I need to do is set the permissions so that new files inherit from the parent directory, but so far nothing I've found via google actually works for me.

Does anyone have any suggestions which might work?

I've got a 750GB drive which was in a unionfs pool, I removed it from the pool and unmounted it then remounted it (so it is definitely not still referenced or in use), then tried to set the a quota on it, but it fails with:

Failed to execute command 'export LANG=C; omv-mkconf quota 2>&1': quotacheck: Cannot remount filesystem mounted on /media/fdc0ff6a-6f4c-41b5-97af-fe647920fefa read-only so counted values might not be right. Please stop all programs writing to filesystem or use -m flag to force checking.

I had the same issue with a RAID5 array, eventually managing to get it to work by removing ALL references to the file system, rebooting and setting quota immediately after boot. A reboot is not getting me going in this case however.

Could someone please let me know the best way around this while keeping the OMV config files consistent with the actual quota config?

Thanks.

Edit: the mount point referenced in the error is not the one I'm trying to set the quota on, it's RAID5 I previously had to remove all the references to set it. An option here might be to manually remove the quota on this file system because it is no longer needed, then I may be able to set the quota on the disk I am trying to use.

Edit2: OMV let me remove the quota on my RAID5 in the web UI, then I was able to set it on the new 750GB disk. My problem is solved, but it might be a little buggy here.

Trim does not seem to be enabled by default, I have no discard option set in my fstab (I run my NAS from a 120GB OCZ Trion 100). Smartctl doesn't seem to report TBW on my system, but the erase count attribute is still 0 after 2700 hours power on time, that means I haven't even generated half the capacity worth of writes (60GB), at which time this attribute should change to 1. I think you would be unlikely to ever see write endurance limits reached either through log files being written or write amplification from lack of hardware trim especially when only 2.1GB of 120GB is used. I think the oldest SSD I own which is now in my HTPC only has about 7 TBW in 5 or so years.

For your four drive non redundant array consider unionfs. It is like a virtual mount point that merges the contents of the disks together, and writes to disks based on an algorithm (such as which drive has the most free space). If a drive fails or you pull it out, you lose only the data that was on it but you don't have deal with having a heap of different mount points and shares. You can easily add or remove drives as you please, so it's a lot better than LVM.

I was copying from OMV to SSD in desktop and SSD in laptop, from two different volumes (3x WD Green in RAID 5 for one and 1x 3TB Seagate for the other), usage graph in OMV peaked at 997mbps. I believe because I'm copying to two different machines on the same switch I should get up to 2gbps assuming no other bottlenecks, but not so. Must be something wrong elsewhere. The bond failover works at least, I can unplug either cable and it doesn't go down.

Booting without the cable connected didn't work. I mounted the disk from a USB bootable linux fix it.

3 hours later after finally getting 802.11ad working, I tried copying from my laptop and desktop simultaneously and I'm only getting 1gbps anyway!

My OMV 2.0 machine has onboard LAN and an Intel PCI-E server NIC. I plugged in the onboard LAN, configured it to DHCP, then deleted the interface on my Intel NIC then created an 802.11ad bond on it, now my machine won't boot past the point where it requests DHCP. See attached photo.

I can't boot into recovery mode either. It responds to ping, but it never finishes booting (so far).

Some help would be really appreciated.

Edit: I just found this which I think is my problem, when I configured the bond I set IPv6 DHCP to auto. Is there no way to recover this?

Edit2: Easy fix, just remove IPv6 stuff from /etc/network/interfaces. Just need a way to mount disk e.g. Ubuntu live CD or any whatever is convenient.

Hi all,

This is my first OMV build, and so far I'm loving the flexibility. I had all my storage in my HTPC, but I didn't like the noise of that always running in the lounge. I also had a web server and mumble server running on a Raspberry Pi, and L2TP VPN on my router, so I wanted to consolidate that. OMV fit the bill perfectly.

My specs are

i3 6100 Skylate 3.7ghz dual core with HT
8GB DDR4 RAM
Gigabyte H170 M-ATX motherboard
(Note on motherboard: I usually don't like Gigabyte, but it was one of the few locally available with 16x and 4x PCI-E slots, rather than 16x and two 1x, and had Intel NIC. I wanted two slots at least 2x for NIC and SATA controller)
HP Intel dual port server NIC (spare from work)
OCZ Trion 100 120GB SSD for boot (crap, but hopefully reliable enough)
3x 2TB WD Green in MDRAID5 - important stuff
1x 3TB Seagate and 1x 750GB Samsung in LVM span - not important stuff
Cheap Silverstone case with 5x 3.5 inch and Lian Li 2x 5.25 inch to 3x 3.5 inch bay adapter (total eight 3.5 inch bays)
OMV 2.0 with backport kernel

I am using it for file/media server to Windows clients, OpenVPN, video surveillance storage, torrenting (100/20 fibre torrenting directly to NAS ) and remote backups for my parents data.

I put structured cabling my house and am getting 113MB/s to/from my Windows 10 machines. This box lives in my linen cupboard next to my fibre ONT, router, switch etc. It used 41w idle with 4 drives spinning and the CPU idles at barely more than ambient temp.

The only issue I had with setup was the onboard NIC wouldn't work with the 3.16 kernel, so I put the server NIC in as a temporary measure but I think it is better than onboard anyway so it will stay.

Here's hoping for many years of trouble free service!

A gbe connection is always full duplex, so send and rec is using different pairs.

Perhaps your cable is bad?

Thanks. I was totally new to OMV until my crash course till 2 am the morning after I built the NAS, so I didn't know about the backport kernel. I'm not really one to read instructions until I get stuck.

I wasn't actually editing settings.json with a text editor, just copying the file from one place to another so the correct config was in the correct place to be loaded when transmission-daemon is started.

The symlink is not present on my system, so that is probably all I need to correct before the web GUI can configure the settings properly.

Thanks again.

I have installed OMV 3.0 on a new build (I had to use 3.0 because it's Skylake platform and 2.0 kernel didn't support my onboard NIC).

When I make changes to Transmission using the GUI it writes to a /etc/transmission-daemon/settings.json, but the actual daemon when it runs loads it's config from /var/lib/transmission-daemon/settings.json.

It's not a big deal because I just stop the daemon with service transmission-daemon stop, then cp the config file into the correct place and start the service again, but I would like to fix it.

If someone knows how I can make Transmission load the correct file that would be much appreciated. Thanks.