Beiträge von Swampy

I'm having the same problem but have a few additional comments:

• I'm waiting for a Time Machine backup to complete. Then I'll try rebooting between steps. Hopefully this will help clean up some of the mess.
• Even though PMS is uninstalled, the sidebar still shows "Plex Media Server" in Sevices.
• Whenever I click this entry, the system throws an error message: "Failed to execute XPath query <cr><lf> //services/plexmediaserver."
• Ykzel is going to reinstall OMV. For me this is really unacceptable. On my system OMV sits on an internal USB flash drive, and it has to be removed to reinstall OMV because the system can't boot from a drive with an install iso while the system drive is present. Just reinstalling OMV is a 2-3 hour process since the ~50 lb. NAS system has to be removed from the media cabinet and moved elsewhere, a terminal & keyboard attached, the system drive removed, and OMV reinstalled. Then OMV has to be reconfigured and the NAS system returned to the media cabinet. And there's no guarantee the PMS plugin will behave any differently after OMV is reinstalled.
• When OMV is reinstalled, what happens to existing data on the system's hard drives? Are the data lost, or do they stay on the drives?
• Is reinstalling OMV a common solution for problems like this? If so, is there anyway to reinstall by changing the BIOS settings and not having to remove the internal system flash drive?
• Can one reinstall OMV on a headless system, or does one have to connect monitor & keyboard?
• Is the plugin really worth all this trouble versus just installing PMS under Debian through the CLI?
• I'm a Plex Pass member and want to use the Plex Pass version of PMS. Considering this, is the plugin really worth it?
• Lekim36: when you mention installing PMS "manually," do you mean a Debian install outside OMV?
• Lekim36: From CLI on my system, running "ls -l" while in /var/lib only returns an entry for "postfix." When you say you "recursively update the parent folder," exactly what parent folder do you mean: /var/lib, /var/lib/plexmediaserver, or another?
• There's no "/var/log/daemon.log" on my system.
• Suppose I went the route of installing PMS under Debian and without using the plugin. What's the recommended way to clean up things in OMV, like the Services > Plex Media Server entry?

Please forgive me, I'm new to BOTH OMV & this forum editor. (I wanted to quote just a few lines but couldn't figure out how. I also haven't figured out yet how to insert comments after partial quotes. So comments are at the bottom.)

Zitat von ryecoaaron

YES it does. Once you enable user home directories in the access rights section, it sets the home directory in /etc/password. This is the same place that samba uses to direct the home share to.

Don't confuse samba home share name and omv home shared folder name. They have nothing to do with each other by name. The both point to the user's folder in /etc/passwd.

NO NO NO

Yes, I have confirmed this in my previous posts. I don't know why you don't believe me. The ONLY time /home on OS drive would ever be used is if you manually create a user. If you enable home directories in OMV and samba and then create the users, you won't have any problems. Do I really have to waste my time creating a video to prove this?

Thought Apple dropped Samba years ago. So I wasn't even thinking of Samba. I thought Apple developed its own version of SMB from scratch over the past 5 years.

I believe you, I just don't understand everything you mean. If I understand you correctly now, you're saying that the "/home" directory I see when I use SMB to connect to the OMV server is (almost?) identical to the "Users" share I also see. "Almost" because /home is a directory on the share, and the share contains "/home."

Let me test my understanding with two questions based on your last comment. Suppose I enable home directories in OMV and create user "Kernighan." Now Kernighan knows *nix pretty well, so he also connects to the server from a terminal window on his Mac and just logs in at the login raster. Where is $HOME in his shell? If I understand you correctly now, because OMV put his home directory into /etc/passwd, his /home directory as a pure *nix user ($HOME or ~) is located on the Users share that was created under OMV and initialized as the home directory through Access Rights Management.

Also, if he issues the command line command "pwd", how will the system respond? Guess #1: the system will respond with /home/Kernighan (this would be consistent with my interpretation that the system mapped /home to the Users share). Guess #2: the system will respond with a different path (not sure what) because /home still lives on the system drive, but the last directory in the path would be Kernighan.

P.S. While a video would be nice, please save your videographical talents for the really tough questions.

Thanks to your help, I finally got the SMB protocol working.

Looking at the results though, it appears that "Enable user home directories" in the SMB/CIFS section DOES NOT refer to the directories configured as user home directories in the Access Rights Management section. I say this because I configured the "home directories" in the Access Rights Management section to put these directories on the "Users" share, but using SMB to access the NAS shows this location AND a /home location, which appears to be /home on the system drive.

From this I infer that enabling user home directories in the SMB/CIFS section means users accessing the NAS via SMB will have access to /home on the system drive, and that this is entirely independent of whether or not user home directories are enabled in the Access Rights Management section.

Can someone confirm this?

This may have an obvious, easy answer, but I'm new to OMV and for me router configuration -- especially port forwarding -- is something of a black art. So please forgive me if the solution is obvious.

The instructions for configuring the Plex Media Server (PMS) plugin at Plex Media Server > Setup Info says:

Forward the following ports from your router to OMV. These ports are used for remote access to your Plex Media Server.

• TCP 32400 (for access to Plex Media Server) [required]
• TCP 32443 (for SSL access to Plex Media Server) [optional]

But my network already has a PMS running on another device (a Netgear ReadyNAS), and I cannot disable it until the OMV server is completely configured and up and running satisfactorily. (There's also a PMS on an iMac, but it's usually off because it's only used for testing.)

The "Manual Port Forwards for Multiple Servers" section of Plex's "Troubleshooting Remote Access" page suggests using external (WAN) ports 32401 and 32402 for two different servers and forwarding them both to internal (LAN) port 32400. But because this might break the existing setup, before doing this I want to be certain I'm doing things right. Here are some things to consider:

• Eventually I'll use the Plex Pass version of PMS.
• Eventually I'll decommission the ReadyNAS, at which time the OMV server should revert to external port 32400
• Currently there's no sign of life for the PMS running on the OMV machine. It's enabled, but the Plex Web Client can't connect to the server, and the other Plex clients on the network don't see the OMV server.
  
  
  • This makes sense, since Port 32400 is not yet assigned to the OMV machine. But before doing so, I need confidence in a roadmap for getting to the final state.

Could someone please provide such a step-by-step roadmap?

Thanks.

Thanks R & D. This is very helpful. I'll comment on general points later, but right now I still have one question.

Now that I understand Rye's #4, to hold user data I'm going to add a share to my server, with users' home directories located on the share. I'm inclined to call the share "Users," but if OMV already has its own naming conventions, I'll use them instead.

So, Debian conforms to FHS file-naming conventions, but beyond these does OMV have its own conventions for shares and/or other things?

Thanks, rye. This is very helpful.

Now what about home directories? Here are some of my questions:

• Are OMV home directories also Debian home directories?
• When a user is added, by default does Debian create a home directory for the user?
• Can one configure the system so that /home does not reside on the system drive? How?
• Does OMV admin have to create a share and then locate OMV user home directories on the share before OMV users can have home directories?
• When other component of OMV mention home directories, as in the example of SMB described above, are these the same as OMV's user home directories or are they home directories solely for the purpose of the component? If the latter, where are they located and how does one control their location?
• Where is this stuff documented?

Zitat von DHGE

BUT: in a private use setting I expect the user homes to reside on the client machines
On a NAS I can live with users without home-dirs.

Not sure I agree with your assumption. For example, my wife uses a cheepo Windoze PC with only 32 GB of mass storage. She's got chronic space issues, and one use for a NAS could be giving her more storage for her documents. (Better than SD cards or USB drives, which frequently get lost, forgotten, or broken.) If so, she would have something resembling a home directory on the NAS drive.

Technically this might not be a true home directory in the sense that she can log into the PC without being connected to the network. Let's call her "true" home directory under Windows 10: C:\Users\<username>. But because of the space limitation, everything is configured to store on the NAS, and user files there are organized under D:\Users\<username>. Since a user logging in on the PC doesn't start with this directory, it's not really a home directory from the vantage point of the PC. But if a PC user logs into the NAS machine as a Unix user, wouldn't one hope /home/username and D:\Users\<username> point to the same place in the file system?

In fact, IIRC, early Sun Microsystem computers did use an architecture with a single heavy-duty server connected to many lightweight workstations. I believe all files, including home directories, resided on the server.

Your comment about /home on the system drive raises a good point: are OMV "users" also Debian "users"? If there's only one "lonesome" user, the answer is "no." Presumably there would be 2 users in /etc/passwd: root and OMD's admin, but after that all other users could be known only to OMV and therefore are not in /etc/passwd.

But again, why restrict the NAS device to being just a NAS? In the 1970s and 1980s I certainly used mult-user Unix systems. Many of them doubled as DNS, mail servers, etc., but personally, I used these machines mainly for programming and typesetting (with *roff), applications that require user-specific storage. These were pre-HTML machines, so no http whatsoever. We used dumb terminals, and we liked it!.

Distinguishing between "private" and "business" settings reminds me of the 1960s distinctions between "scientific" and "business" computers before IBM's System 360 computers came along and of the distinction between "scientific" (FORTRAN) and "business" (COBOL) programming languages before more flexible and capable languages (PL/I -> C) came along. At any point in time technological limitations may require distinguishing between different application realms, but Unix (and Novell) proved it's always a bad idea to cement these differences in fundamental architectures tied to specific applications.

But enough philosophy. My question is just looking for something broader. I want an overview, explaining from 30,000 ft up, how OMV organizes user-specific storage areas and then, on the ground, explaining exactly what "user home directory" means in each part of OMV, how it's used, and how it relates to storage locations on the Unix host system.

IMHO, people really shouldn't have to go around hacking into Unix system configuration files to try and guess how OMV interacts with the underlying OS or what "User home directory" really means in varied contexts in OMD. Indeed, I wouldn't be surprised in a substantial majority of new OMV users never even heard of /etc/passwd, let alone have ever looked inside one. And if they hadn't, they might guess that /etc/passwd has something to do with usernames and passwords, but why would they even think it has information about home directories?

IIRC, Unix files actually consist of linked lists of inodes. So technically it may be possible to figure out all this stuff by experimenting with different configurations and reverse engineering what's actually going on by examining the inode lists. But just because this is possible doesn't mean there shouldn't be an easier way: one that explains in plain English (or whatever) what home directories are and how they're used on OMV systems.

Thanks. You bring up several subjects with which I have limited familiarity. Furthermore, I thought I'd have my server completely configured and running by devoting the entire weekend to the project. This turned out to be wishful thinking, and going forward I'm going to have to tackle one task at a time. So I'll probably get back with more questions and results, but this will not be immediately.

My $0.02 as an OMV 3.x/Debian beginner.

Can someone please explain "User home directory" (UHD) or direct me (us) to good, general documentation about the subject? I see this term in various places in the GUI, but I'm uncertain what it means.

For example, Access Rights Management" > User > Settings wants a location for the UHDs, but the drop-down menu only shows "None" and the names of the two shares I've set up ("Media" and "Zotero_Library"). These are both special-purpose, and neither is an appropriate place for user home directories. So does "None" actually mean no UHDs will be created -- thereby negating the "Enable" switch, or does it mean they're created in some unspecified default location, or does this imply a need for a third share (e.g., "Users") where users will have their own home directories?

To continue, wherever one sets up the UHDs, is this the same location for home directories as one would get if one logs into *nix and a shell (i.e., "~")? E.g., on MacOS computers home directories are: /Users/<user name>, but Debian documentation implies: /home/<user name>. In either case, is there a default drive when using OMV? Both MacOS and Debian are primarily designed to be computer OS's, so they are likely to assume a large "system" drive is the location of both the OS and the user part of the file system. But OMV wants to keep the system drive separate from user data, and using a small USB drive or SSD is common for this purpose. So one would assume that /home is not on the system drive. If not, then where is it by either default or convention? If it is, then isn't this silly because OMV doesn't want user data on the system drive?

Following this line of thinking, is there a way to explicitly locate /home on a different file system than the one where the OS resides? The Filesystem Hierarchy Standard (FHS), to which Debian subscribes, implies that this is common. But the OMV ISO installs Debian before installing OMV, so it would seem that by the time a sysadmin accesses the OMV GUI the location has been decided. But where? Also, I think most people would like consistency, so that their home directory is the same no matter how they access it (CLI, OMV GUI, connecting device's file app, etc.). Is this actually the case?

But whatever the story is about user home directories, the picture becomes even more muddy as one uses other parts of the OMV system. For example, Services > SMB/CIFS also has an "Enable user home directories," but this one does not have a text box or drop-down menu to choose a location. What does this mean? Does it mean the home directory is unique for SMB access and located in some unspecified location? Does it mean the SMB/CIFS subsystem uses the settings established in Access Rights Management > User? Does it mean /home/<user name>?

There are lots more questions like these, but after searching much of yesterday and this morning for a clear discussion of the general topic, I'm giving up on finding one. Instead, I'm hoping someone here can either direct me (and others) to such a discussion or give a clear summary here. Pleeeze.

P.S. The FHS documentation is an excellent example of documenting filesystem conventions. Something similar for OMV -- addressing things that are unique to OMV and the specifics of how OMV handles the more general *nix conventions -- would be really helpful.

But even if I bought a cheap domain, wouldn't I still have to pass Port 80 to the OMV NAS? The ReadyNAS is still using Port 80.

I'm confused about how to proceed setting up security with my OMV NAS and hope some of you can help me.

What's Happened Until Now

A Netgear ReadyNAS at home is being retired and replaced with a DYI NAS running OMV 3.x. The OMV NAS is up and running, and now I'm configuring security. The current step is making it require secure connections only.

My router is a Linksys WRT1900ACS. Eventually I plan to change it over to Open WRT, but not until the OMV NAS is completely stable. I'm not going to start messing with the router until I'm finished setting up the NAS.

Initially I created a self-signed certificate in OMV and then set the system to require secure connections only. This didn't work because browsers like Chrome and Firefox would not connect because the certificate did not come from a recognized source.

So I installed the LetsEncrypt plugin and tried to make a recognized certificate. This didn't work because the domain name is illegitimate. When I originally configured OMV, I used the default domain name, "local", both because it is the default and because I mainly use Apple computers, which also use "local" as the default domain.

My Confusion

I'm confused about what domain name to use and how to use it. Technically, I think we should be calling these things "resource names" because an address like "foo.bar.bas" is really a subdomain of "bar.bas". But both names are interchangeably called "domain names," and part of my confusion is what the documentation is referring to when it says "domain name."

Another part of my confusion is which domain name to use: internal or external. Every device inside my home could easily be in a domain with names like OMVNAS.local, computer1.local, etc. These names can work perfectly well within the network administered by the router. But then again, maybe the domain name needs to be the "external" name used to reach devices from beyond the router. Which one is it?

The router natively supports "external" domain names from either of two DDNS providers: No-IP.com or Dyn.com. Presently, the ReadyNAS can be reached by using one of two names provided by No-IP.com. To reinforce my earlier point, both end in two-part higher-level domain names over which I have no choice. E.g., foo.ddns.net. I can only choose the first part (foo), and my current plan allows at most 3 fully qualified domain names. Moreover, the ReadyNAS takes up to two Ethernet wires and therefore is already configured to use the two names I'm using. And perhaps more important, the router already maps ports 80 and 443 to these two connections.

I can't use the two existing names because they're already taken by the ReadyNAS. I can't add a third name because OMV wants port 80. So what's the best way to proceed to make the OMV NAS require secure (SSL/TLS/HTTPS) connections?

Thanks. I had just came back here to say I'd solved the problem on my own. For the benefit of others, here's what I did:

• Initially I had booted up with just the installation drive in a usb 2.0 port. Then, once the installation got to the point of listing Partition Disks, I backed up several steps to the Detect Disks step, inserted the OS usb into a 3.0 port, and retraced the steps up to partitioning the system disk. This did not work: when the installation script got to the select Partition Disks step, it only gave an error message saying there was no root partition.
• So I rebooted the entire system with only the installation drive in a usb 2.0 port. But this time when the installation script paused for the first user input (Select Language, I think), I inserted the OS usb into an external usb 3.0 port and continued with the installation. This worked up to the point where the system tries to reboot from the newly created OS drive. I plan to finish the installation tonight.

Disconnecting the SATA drives would have been my next step, but I was very glad I didn't have to do this. Disconnecting them would have been a real pain. I'll see what happens tonight. Thanks for the help!

Hi,

I am struggling to install OMV 3.x from openmediavault_3.0.86-amd64.iso onto a USB drive, which will become the boot device for my DIY NAS. But when the install process gets to the point of selecting a drive onto which OMV will be installed, at the Partition Disks step of the installation, the USB drive does not appear in the list. What should I do?

Here's some background information:

Hardware

• Motherboard: Supermicro X11SSH-F-O
• Future boot USB drive: 16GB MX-ES brand SLC USB3.0 drive, currently formatted as FAT 32. VERY IMPORTANT: This drive will occupy USB Port 10 on the motherboard, which is a completely internal USB port designed for housing a boot USB.
• RAM: 16 GB (2 x Crucial 8 GB DDR 4 2400 ECC #CT2K8G4WFS824A)
• Other mass storage: 3 x 4TB WD Red SATA drives (12 TB total)

Experience So Far

• Initially I created an installation boot USB by copying the install iso to a USB 2.0 drive. I did this on one of my Macs using the dd command in CLI.
• I then tried to install OMV from this drive. First I inserted the drive into one of the system's external USB 2.0 ports. Then I modified the BIOS to make USB the first choice for booting. Then I booted up the system. But after initial boot up, it just showed a blank screen and never moved from it.
• Then I created another installation boot USB on another device. Only this time, instead of using the CLI dd command, I used the unetbootin app. Previously I had tried unsuccessfully to use this app on an iMac running OS X Yosemite; the program never initialized its main window properly. This is why I used the dd command instead. But this time I installed unetbootin on a MacBook running MacOS Sierra, and the program worked.
• Nonetheless, when I used this installation USB drive instead of the original one, I got the same results as before: nothing but a blank screen.
• Up until this point I had been assuming the MB would look at available USB drives and choose the one that was bootable. Now I began to think that maybe the system was just confused by two USB drives being present. So I removed the future boot USB drive from USB Port 10 and rebooted. This time the installation actually began successfully.
  
  
  • Because of this, I believe the earlier problem was not due to using dd instead of unetbootin; instead, the problem was due to having a second USB drive present at boot up.
• But now, when I get to the Partition Disks step, the system only lists the SATA drives. Of course: initially they are the only target locations available. So I go a few steps back in the installation procedure, insert the future boot USB in Port 10, and continue the process. The drive still does not appear among the Partition Disk choices. This is where I am stuck.

What should I do next?

Please advise on using OMV to receive files attached to emails and store them separately as files in a temporary storage area on a NAS. A more extensive explanation follows.

Last winter I built a DIY NAS server, planning to use OMV as its OS. But due to some personal issues, I'm only now getting around to installing an OS & configuring it. Originally I planned to use the NAS only as a media server for Plex & Kodi and to house a digital library of pdf files. I already have an older NAS (a Netgear ReadyNAS), although it is running out of space and can't run recent versions of Plex.

Currently, to access the library from a computer, tablet, or smart phone, one only needs to establish a WebDAV connection to the library's share. To maintain the library, I use Zotero with the Zotfile plugin. This works fantastically, and I intend to continue this on the new NAS.

Previously, to add items to the digital library, if I did not already have a pdf, I would create a pdf using a departmental scanner at work. The scanner would save the pdf file on a mass storage device (typically on a computer or in Google Drive). Then using Adobe Acrobat Pro, I'd make the pdf searchable with OCR and reduce its size. I'd then simply link the pdf file to its bibliographic entry in Zotero and let Zotfile do its magic: rename the file and move it to the appropriate place on the NAS drive.

But earlier this summer the department replaced and reconfigured its scanner. Instead of scanning and storing files directly on a mass storage device, the new system can only send scans as attachments to email. This creates two serious problems. First, it adds the extra steps of opening the email, finding the attachment, and saving it as a file in mass storage. Second, and even worse, email servers have (configurable) limits on how large attachments can be. Typically, 30 MB or so is the maximum. This means that most color scans are too big, and the only way to save such a scan is to break the original into smaller pieces, scan each piece, save the separate files attached to different email messages, and recombine them into a single document.

For example, using my home scanner, which can write directly to drive space, I did a test with a 36-page journal article marked up with colored highlighting. The initial file was 41MB, 31 MB after OCR, and 7 MB after reduction. But the email server rejects the large original sent directly from the departmental scanner. So I had to divide the original into multiple parts, scan each separately, open each separate email, save each part to disk, and recombine them. Only then could I run OCR on the reassembled document, reduce it, and let Zotfile store the 7 MB file on my NAS.

In general, I need to use the departmental scanner because it is much faster and more capable than my little scanner at home. I'm in the middle of digitizing my library, and soon will start scanning much larger, color documents. Because of the added steps doing this peacemeal through email, using the new system this will be much more time-consuming than necessary. So I'm thinking of configuring the new NAS as a mail server (with either no size limit or a very large limit for attachments) and using a Perl or PHP script as described here, first to check that the sender is the scanner and then to save the attachment in a temporary storage area. This would get around the email limitations and allow Zotero/Zotfile to attach each pdf to its bibliographic record and to store the pdf in the library without all the unnecessary intermediate steps.

Nonetheless, I'm a bit wary of this because I don't want to open up the server to security risks by adding mail service to its functions. Perhaps some kind of FTP service would facilitate the email file transfer more securely.

So my questions pertain to this email/ftp service. Is something like this available for OMV? Would it conflict with the server's main functions as media server and digital library storage? Are there other, better solutions than adding email (smtp) to the server's functions?

Thanks.

For a OMV newbie like me, some of this stuff is hard to figure out. E.g., I don't really know what the boot drive does. Does it just hold the operating system, or is it used for swap space too? Can one install apps like Plex on the drive? Etc.

So, if I take your advice and go with a USB. Is USB 2.0 good enough, or should it be a 3.0 drive?

How big should it be?

I've been looking at Sandisk Extreme Pro 16GB, which are going for as little as under $15. But the description says it's encrypted, and I'm not sure this is a good thing for a boot drive. Your thoughts?

I'm putting the finishing touches on my first DIY NAS system and need some advice about choosing a boot device. The motherboard is a Supermicro X11SSH-F, which has a M.2 slot that takes a 2280 card and uses PCI-E 3.0 x2. It seems to me the best boot device would use this slot, but I'm concerned about compatibility. Supermicro has a reputation for being a bit finicky about memory, etc., and the only M.2 devices Supermicro lists as having been tested with this board are by Toshiba and start at 256GB. Because of budget & need, I am thinking more along the lines of 128GB, which is still probably much more than I need. I've been looking at the _{128GB Intel 600p for $65} and the _{Samsung PM951 for $81} as possible choices.

But I worry about compatibility with OMV and the motherboard. As I understand it, the OS on the M.2 drive has to support UEFI. Also, as I understand it, OMV wants the entire drive to itself, so paying this much for 128GB of memory may be a waste.

One option would be getting a less expensive 120GB SATA drive. This PNY drive costs only $40, and there are several other choices in this price range.

The other option would be using a USB drive. These come in smaller sizes and therefore would save money. But I've read that OMV will wear out a USB quickly.

So what do you folks with more experience recommend?

Thanks, Ryecoaaron & Subzero79!

OK. So let's focus solely on functionality. Forgive me because I'm trying to make sure I understand this. Below, where I'm able to piece together information to guess at an answer, I put the guess at the end in square brackets, "[ ]."

Using the WebDAV plugin is it possible:

• For individual users to make one or more of their own folders accessible via WebDAV? [If a user, say Jon, is a member of a group with WebDAV access, the user can log in and then create folders and files. Jon will be the owner. -- But what if the folders/files are created some other way, say by uploading with FTP? Can Jon subsequently make them accessible via WebDAV?]
• For the owner of a WebDAV share, Share A, to give a subset of all users (e.g., Group "Stark") rwx access via WebDAV to a folder and its offspring? [The owner, Jon, can accomplish this by setting *nix group to Stark and permission to 770. But see #3 & 4 below]
• For the same owner of the same WebDAV share to give another subset (e.g., Group "Lannister") only rx access to the same folders? [The owner, Jon, can accomplish this by setting *nix group to Lannister and permission to 750. But will changing the group this way break #2?]
• For the same owner of the same WebDAV share to deny all access to these same folders to a third subset of all users (e.g., Group "Targaryen")? [The owner, Jon, can accomplish this by setting *nix group to Targaryen and permission to 700. But will changing the group this way break #2 & #3?]
• To have the same kinds of access or non-access as described in #2-4 for a different WebDAV share, Share B, only this time the group assignments would be different? E.g., Members of Group Stark have complete rwx access to Share A (#2 above) but no access to Share B, while members of Group Targaryen have complete rwx access to Share B but no access to Share A.
• To have the same kinds of access or non-access described in #2-4 for a third WebDAV share, Share C, only this time the access applies to different groups, and individual users previously in different groups are now in the same ones, while users previously in the same group are now in different ones? E.g., Rob and Jon are both in the Stark group, Jamie and Cersei are in the Lannister group, and Aemon and Danni are in the Targaryen group. But Aemon and Jon are also in the "Black" group, Danni, Rob, and Cersei are in the "Crowned" group, and Jamie is in the "Handless" group. While still preserving access to Shares A & B, is it possible to give the Black group rwx access, the Crowned group rx access, and the Handless group no access to Share C?
• To change access (via group) on the fly. E.g., if membership in a WebDAV group merely allows login access, if Jon logs in as a member of Group Stark, to exercise his rights as a member of Group Black, does he have to logout and login again as a member of Black, or can he simply chgrp after logging in?
• To change access without reconfiguring groups. E.g., suppose we realize that Jon should be in both the Stark and Targaryen groups. Do we now have to completely revamp what we did before (e.g., creating a "Stark-Targaryen" group for Jon alone), or is there some other way to assign Jon to both groups and still give him rwx access to Shares A, B, & C?

Wow! Thanks Niemand and Luxflow. Now I understand that OMV adds its own file system layer on top of the standard *nix ugo file system. And, if I understand luxflow correctly, the WebDAV plugin bypasses the OMV layer (by using NFS?).

I only have three comments/questions in response.

• Subzero79's document says, "these privileges have absolutely no effect in the core file system layer where they belong (let that be ext3, ext4, xfs, or jfs). Every change you do in privileges section, OMV will not execute a single chmod or a chown over that folder. For general knowledge purposes every share created under OMV webUI is created with default 2775 permission, with a root:users folder ownership." I'm not sure what to make of this. Because the privileges "have absolutely no effect" on the *nix file system, does the part about "default 2775 permission, with a root:users folder ownership" exist solely in the OMV file system space, so that in the underlying *nix the owner could simultaneously be foo, and the group could be bar, with chmod +700? OR, is the part about the default permission and root:users the exception to the earlier rule, so that creating a share in OMV actually does have an effect on the core file system layer, namely creating a directory structure with root as the owner and users as the group, with permissions 2775?
• I've used WebDAV on a number of different systems, including among others a Netgear ReadyNAS that I administer myself and a Sakai system on which I am a user. If you're unfamiliar with Sakai, it's an open-source learning management system used at many colleges & universities. As such, it's organized around courses and projects, each of which has a "Resource" storage area. Sakai apparently generates a WebDAV URI for every separate "Resource." IIRC, it actually generates two: one with write permissions and one with only read permissions. Since Sakai is open source, I wonder how hard it would be to take the relevant code from Sakai and modify it for OMV, so that when a share is created it can easily be accessed via WebDAV. Also, I'm not sure of the details, but I don't think this scheme would necessarily require a specific owner and group, unless OMV breaks when it's not using root:users.
• Awasu has an article on managing permissions in OMV, including a shell script to fix ownership and permissions. This seems to take care of the native *nix side of the equation, and it would fix things if the WebDAV plugin, NFS, or some other vandal mucks things up. But what about the other side? Suppose a share already has the proper *nix ownership & permissions, and now all one wants to do is to fix the corresponding OMV settings. Are there shell commands that can do this? And if so, would implementing a schedule for running the script, as the article recommends, be all that's necessary to overcome the plugin's bypassing of the OMV file system layer?

Zitat von luxflow

This plugin has limit
It cannot give shared folder per user
All webdav user share have access to same shared folder

I don't understand this completely. Here are my questions:

• Is this the plugin's only limit over more complete implementations of WebDAV?
• When you say "shared folder per user," do you mean it doesn't allow (a) individual users to set up their own folders accessible via WebDAV, (b) owned by the individual user, and (c) shared by the users who have access to the folder?
• When you say, "All webdav user share have access to the same shared folder," do you mean there can only be at most one WebDAV folder on the server and anyone who uses WebDAV has +rw access to all the files within it?
• Do restrictions on a top-level folder accessible via WebDAV apply to all child folders underneath this parent folder?
• Since OMV is based on Debian, would a WebDAV implementation for Debian, and bypassing OMV itself, be a more complete, trouble-free way to implement WebDAV.

Zitat von gderf

Just because all those fans "come with it" doesn't mean you need to run them all.

Great point!

Here's my thinking about this:

• There's no way to know ahead of time how much cooling this rig will need because the actual amount of heat generated will depend on ambient temperature, usage, and components, while cooling will depend on the specific aerodynamics and thermodynamics of the case, the air flow of the installed fans, etc.
• Nonetheless, I am working on a budget. So if I have to purchase fans eventually, I need to try to include them in the budget now. This means estimating the fans I need to the best of my ability.
• The Rosewill case I plan to use can take up to six 120mm fans (3 intake & 3 motherboard cooling) plus two 80mm fans (for exhaust).
• Additionally, there's the CPU cooler. So the system could have as many as nine fans!
• Rosewill sells several variants of the case, but all of them ship with fans. The model with the fewest fans comes with only 2x120 mm intake fans and 2x80mm exhaust fans.
• So until I can determine otherwise, I assume that a minimum of four fans is necessary to cool this case properly.
• So I need to plan on using at least four fans, 2x80mm exhaust and 2x120mm intake.
• Adding the CPU cooler, this brings the initial total number of fans to five, which is exactly the number of fan headers on the X11SSH-LN4F.
• Several user reviews of the Rosewill case have said the stock, included fans are noisy.
• Silence is an important design constraint for this NAS.
• Besides, the included fans are not PWM.
• So the budget should include four replacement fans for the four included ones.
• Nonetheless, what I'll probably do is build the server with the stock, included fans and then see how much noise they make and if they keep the system cool.
• Afterwards, I'll order at least four quieter PWM fans, and possibly more if the original fans don't keep the system cool.