These days many popular applications permit 3rd-party add-ons, a.k.a., plug-ins, modules, components, etc, and people are fairly comfortable installing them.  There are also several community-created unRAID add-ons as well.  If you have been following the forum, then you know that a goal of the next major release of unRAID Server will be to include a "formalized" method for creating add-ons.

 

Why has this not been an emphasis sooner?  The reason is, it's one thing to install a Firefox add-on and then discover something is not working right - you can just remove the add-on or re-install Firefox.  But with storage, there may be a possibility of data loss.  The first rule anyone in the storage business must follow is: "Don't lose the customer's data!".  This is a major reason why the unRAID Server release process is, um, methodical - I don't want to release anything that could possibly corrupt user data.

 

So I anticipate users will be, and should be, more wary of installing add-ons to a storage platform.

 

As always, comments welcome.

 

I am very interested in seeing the results of this poll...

 

 

Very comfortable.

 

But not only am I unRAID club hacker, I'm a client too. 

 

 

I am fairly comfortable doing the unRAID addon's that the community has put together.

 

I currently run BubbaRaid, unmenu, and some of the scripts that have been created.

 

I love the idea that version 5 of unRAID is going to have some stuff that will make it easier for things like this to happen and I am looking forward to seeing how it all turns out.

This community is very helpfull and I could say I am comfortable installing add-ons.

 

I am very sure that there wil be a standard procedure on how to do it.

 

Its worth keeping in mind that only a tiny percentage of code on unRAID is written by Limetech. This is a guess but its probably a fraction of one percent.

 

This is not to take away from Limetech in any way but it does add some context to the question.

 

Its not about who does the code its about testing and quality control.

Its not about who does the code its about testing and quality control.

 

And who answers support calls and emails, and works with customers to restore their data when something goes wrong

VERY good point. Again i wasnt trying to take away from your work. Its more often than not about skill and vision. I mean Ferrari didnt invent the wheel but they still make Ferraris.

 

But support is something that is vital in this poll. I would never ask Limetech for support for any addon but now you say it I am certain other will just because they dont know the distinction between official and unofficial.

 

I am not sure how this could be handled correctly. You dont want to be supporting random addons but equally you dont want to be telling your customers no i cant help you with this.

 

I am getting OT here but I still think a separate domain that the community can maintain with a big NEON sign saying if you use addons dont ask Limetech for support before asking us for support.

 

Alternatively a tiered support model where Level 1 support could officially be the forum.

Needs: Minimize potential risks.  Minimize support incidents.  Aid support incident resolution by reducing resolution time and skill needed to resolve incident.  Add features that are conducive to third-party support (forum).

 

Some suggestions:

 

Safe mode - away to boot with stock unRAID and ignore all extensions.

 

API Class - Have 2.... one that includes only "safe" commands, and another that will access "dangerous" commands.  Then it is easy to scan code to see if an extension is potentially dangerous, or can be assumed to be safe. (I'm ignoring cloaking of intentional malicious stuff).

 

Logs - extensions must be "announced" in the log when initializing.  Then you can see what on the system in the syslog.

 

Rating system - if they are hosted by Lime, a feedback/rating system can be employed.... and even a sandbox for untested extensions.... move from sandbox to public is like AMO.mozilla.org.

 

Keep the "old" unRAID Management Interface around for at least the transitional versions -- on a different port maybe.  Can also tie into safe mode.

 

 

 

Needs: Minimize potential risks.  Minimize support incidents.  Aid support incident resolution by reducing resolution time and skill needed to resolve incident.  Add features that are conducive to third-party support (forum).

 

Some suggestions:

 

Safe mode - away to boot with stock unRAID and ignore all extensions.

 

API Class - Have 2.... one that includes only "safe" commands, and another that will access "dangerous" commands.  Then it is easy to scan code to see if an extension is potentially dangerous, or can be assumed to be safe. (I'm ignoring cloaking of intentional malicious stuff).

 

Logs - extensions must be "announced" in the log when initializing.  Then you can see what on the system in the syslog.

 

Rating system - if they are hosted by Lime, a feedback/rating system can be employed.... and even a sandbox for untested extensions.... move from sandbox to public is like AMO.mozilla.org.

 

Keep the "old" unRAID Management Interface around for at least the transitional versions -- on a different port maybe.  Can also tie into safe mode.

 

Absolutely fantastic ideas!  Especially the Safe Mode!  That only means an additional boot menu option, right?  An option that would ensure the stock bzroot, bzimage, and go files are used (I *think* that is all that is necessary).  Could that be done now, if someone figures out the correct syslinux.cfg lines?

 

Completely agree about full logging of extensions, and for that matter, it would be good to see most or all of the tweaks to the system logged.  A little peer pressure, and the rating system, could help to enforce that.

It seems to me that addons generally fall into 2 basic categories …

1. They can be added onto stock unRAID

2. They require libraries that require another OS build (like bubbaraid)

 

They can also be categorized in another way ..

1. They have something to do with unRAID (i.e., need unRAID APIs)

2. They have nothing much to do with unRAID, except the desire to collocate them on the unRAID server.

 

I think it would be smart to allow unRAID to be installed in one of two ways – “appliance mode” in which you use the stock unRAID OS distribution; and “installation mode” in which case you would need to install unRAID into an existing Slackware environment.  (In both cases, the license key can still be read off of the USB key and used for license verification).  BubbaRAID would be a special case – kind of a hybrid of the two, and supported in the forums.

 

In appliance mode, you’d be able to run all (or the vast majority) of addons that are integrated with unRAID (like unmenu and other GUI enhancements) and those packages for which library support is provided in the appliance kernel.  You may not be able to run torrent software or other pieces of Linux software for which support is not provided in the “appliance mode” kernel. 

 

In installation mode, you wouldn’t have any limitations.  But obviously installing software app X on the standard Slackware distro is not an unRAID support issue.

 

I think this might be a good way for Tom to address the licenseing and support issues.

 

From a support perspective, there are several things to consider:  Core unRAID functionality, data/system integrity, and extension functionality.

 

Obviously if core unRAID is working fine, and an extension is not working properly or as expected, that has to be shunted to the extension developer.... and I have to hope that most users will do that.

 

If an extension breaks core unRAID functionality without affecting data/system integrity, that should be fixed by booting/enabling unRAID safe mode.... and then shunting it to the developer. 

 

If an extension affects data/system integrity (i.e. causes data loss, configuration loss, or otherwise does something that either hoses the system so it won't be fixed by simply booting into safe mode) that is going to be where the sparks fly.

 

Simply making a "Safe mode" option in syslinux won't cut it, since things added in the go script would still be run.  While the syslinux config is a likely place to implement the safe mode choice, hooks in scripts have to be in place in order to have a true "stock" system boot with a stock go script.... and not have anything else run.  The syslinux config can set a flag or command parameter that can then be checked in scripts to denote that the system booted in safe mode, and then code bracket system and extension scripts to account for that.

I absolutely love the prospect of using my unRAID box for more than just storage functionality.  Most of our systems use less than 5% of their processing power, so lets put it to use.  With that said, I am extremely paranoid about potential data loss and it is a concern when using an add-on that was not produced by the creator.  I like the idea of multiple run-time modes or categories - 1 that is just an application add-on, and 1 that enhances the storage functionality.  I would lean towards not using anything that touches the storage component, and I would like that warm fuzzy feeling of knowing I am running in a mode where nothing can touch the storage component.

 

As a user, I would not hold lime-tech responsible if an add-on crashed the server; however I might if data were to be corrupt or lost because of it.  Data loss, at any cost, is not acceptable for a storage product.  Then again, it is the users choice to install the add-on.  Presenting a user with a EULA or "I Agree" button for installing any add-on's may be enough to convey the potential for problems.

 

The top things I'm looking for with an add-on:

1.  VMWare Server functionality (Easily) - I understand it is available today, but I do not have the time or patience to set it up.  A VM offers a way to install whatever the user wants to without the fear of stepping on the Storage component.

2.  Notification scripts - Temperature and drive failure.  Every other storage product has this ability built in, why not unRAID?

3.  Backup data to USB drive - Again, possible now, but not as easy as it could be.

 

While I do not have much difficulty in the installation, the friends that bought unRAID based on my recommendation rely on me to help them set it up, maintain and trouble-shoot.  Since I moved to another state, it has been difficult to support remotely.  I would love to have a configuration for them that is so easy that a caveman could do it.

 

Regards,  Peter

While I do not have much difficulty in the installation, the friends that bought unRAID based on my recommendation rely on me to help them set it up, maintain and trouble-shoot.  Since I moved to another state, it has been difficult to support remotely.  I would love to have a configuration for them that is so easy that a caveman could do it.

 

Regards,  Peter

 

Thanks for the feedback Peter.  What do you find are the common issues you need to help them with?

The good thing is that they all have the same configuration with the original Asus motherboard and Promise PCI cards.  They are all also running 4.3.3.  The two most frequent issues are:

 

1.  Drives show up periodically as missing so the array does not start.  That scares them the most and befuddles me as well.  So far, the getting them to calm down and just use the mouse to click on the "reboot" button has solved it.  It seems that sometimes drives are not recognized during boot.  I am worried when the day comes when the drive does not come back (inidicating a real failure).  But that will also be a good day in that they can see how recovery works.

 

2.  The more serious issues is when the web page locks up and they are not able to shut down the server anymore.  At first I would walk them through telnet commands (gulp!) to get to the server and then the Linux commands to gracefully shut down the box.  Then I just created a script from one of the posts here on the board.  That has been copied to the flash drive so the pain has been minimized.  They are running 100% vanilla installs (no unMENU or BubbaRaid, etc.) as I want to minimize variables when helping them out.

 

As mentioned I have not moved them to 4.4.2 or above as I sometimes have these issues as well a few others.  So far, 4.3.3 has been the most stable for me and hence for them.

 

Regards,  Peter

Bring 'em on!

 

I tried installing a few add-ons when I initially built my server (BubbaRAID, unMenu), but they didn't work for some reason (something I screwed up, I'm sure).  Since then, I haven't bothered to try them again and I've been running unRAID 4.4 stock.  I do plan on attempting to install some add-ons after I upgrade to 4.5.

 

I would love to see a simpler add-on installation procedure.  Ideally, the user should not have to manually edit any scripts or mess with the files on the flash drive.  A Firefoxish one-button install (and uninstall) or as close to that as possible would make me feel a lot more confident about trying out different add-ons.

 

I also love the idea of a 'safe mode'.

I feel fairly comfortable already installing a few add ons, but I'd feel even better if there was "official" support for doing so. I mostly run additional services on my unRAID (like SABnzbd, HandBrake) versus changing how unRAID works.

It would be nice if you could add things like you do in WHS. Copy the files over, go to add-in's in the control panel & click on install. Couldn't be easier.

 

It's not that difficult in Linux either. But it's one of those things - Use it or lose it. You learn how to install the manual way, but if you don't do it regularly you forget how.. 

It would be nice if you could add things like you do in WHS. Copy the files over, go to add-in's in the control panel & click on install. Couldn't be easier.

 

I'm sure this will occur when 5.x comes out.

 

It's not that difficult in Linux either. But it's one of those things - Use it or lose it. You learn how to install the manual way, but if you don't do it regularly you forget how..  

 

I know what you mean. I always forget where the stuff is in the microsoft menus.  

If their tried and true add-ons, like FEMUR, powerdown, etc... and their is heaps of forum talk and support, I would feel very comfortable installing it. I'd feel uncomfortable using my unRAID as a VMware Server or something like that, though it interests me to read about it. I like to keep my unRAID server storage centric, like you wouldn't use your desktop as a production server.

If they have a great interest to me, I'll try them out. I normally install these on test unraid server that I have, to see the impacts/results. If all goes well, it gets installed on my main server.

I don't think I'd be the one to install FIRST but yes I'd be comfortable.

 

The idea of a plugin system gives me lots of nasty ideas, sadly I'm not a programmer.

I think plugins would continue to promote how simple and useful unRAID is. I have a rather decent background in Linux, but its been a while since I've had to type a command and man I've been rusty lately since I tried this a few days ago and I'm sure Joe L is wondering who this wack job is.

 

The one thing I think would really benifit a plugin system would almost be the same approach this particular forum software uses. Approval. The way SMF aka Simplemachines.org does it they have a team of mod reviewers that takes a mod aka package and installs it or reviews it for security or problems in general before allowing it to go public on their site. You the user then downloads the package and install it. Am I saying that the users here should take their time to do it for free or should the Staff at limetech do it for free? Well NO, but at the sametime I'm sure there are some good volunteers here that would probably donate a bit of time now and then obviously time permitting.

 

How I would approach it is establish a good baseline of existing members here that can look at code and determine more or less what its trying to do when ran. Some Plugins could/would be complicated, which would slow down the approval hours, days, weeks. Heck I've seen mods on SMF take a month or more because it was rather complicated and delayed. After a group is established allow Plugins to be moved or posted in an area in the forum or the site as "Certified" or "Trusted". If need be have a section for "None Certified" or None "Trusted"

 

Am I worrying about something that is probably nothing? Honestly data is data. When its gone its gone and honestly trusting your data to a script or a plugin that is nothing more than "rm -fr *" kinda freaks me out.

 

I guess the only draw back to Certifying a plugin is how you do up the disclaimer. I'm sure limetech doesn't want to assume responsibilty for anybody's lost data. I mean a lost rip of Transformers is nothing compaired to the loss video or pictures of your kids birth or wedding.

 

Just food for thought. I'm sure somebody already has the stick on this or has a better idea, but just throwing it out there.